JP2836486B2 - Multi-dimensional multi-level image compression / expansion method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multi-dimensional multi-level image compression / expansion method, and more particularly, to a pixel correspondence between a compression side and a decompression side such as image transmission between different models, or a frame correspondence on a time axis. The present invention relates to a multi-dimensional multi-level image compression / expansion method capable of performing high-efficiency image expansion in a system where is not guaranteed.

[0002]

2. Description of the Related Art Various methods for compressing image information have been proposed. For example, when the signal level is equally divided for each sample value of the image signal converted into a digital signal, and a linear quantization (equal quantization) unit that replaces a value included in each range with one representative value is employed. In general, when it is assumed that the difference between the representative point and the original value is not known, 6 bits (64 gradations) to 8 bits (256 gradations) are required for a natural image.
In order to record a signal obtained by digitizing an image signal by equal quantization as described above, it is necessary to handle a large amount of information as described above for each sample value.

[0003] Therefore, in order to encode a signal with a smaller amount of information, a portion where the signal changes little is sensitive to the change, and even if there is a certain error in a portion where the signal changes drastically, it cannot be changed. Using the characteristics of human vision and hearing that are difficult to detect, or using the correlation on the spatio-temporal axis in the information signal targeted for recording, for example, decomposing an image into pixels Utilizing the fact that the approximate value of the original information is transmitted using the height of the adjacent correlation between the luminance values of the pixels, the difference between pixels or the difference between frames is transmitted, or the fact that the high frequency component is small is used. The digital data obtained by compressing the data amount by applying various high-efficiency coding schemes that reduce the amount of information per sample by reducing the frequency element Record, Conventionally, transmission, transmission, and reproduction and reception of digital data whose data amount has been compressed as described above, data expansion is performed to restore an image. Is well known.

[0004]

In the above-mentioned conventional compression method of general image information, it is important that restoration of decomposed pixels is performed well. In many cases, the condition is that the number of pixels in an image (expanded image) is the same. Therefore, when a compression / expansion operation is performed between images having different numbers of pixels, interpolation of pixels after expansion is performed separately. However, in the conventional image information compression method, only the true valid information is extracted, and it is not restored. It means that the system is dependent on.

By the way, as described above, two pixels having different numbers of pixels are used.
As an example of a case where the pixel densities of two images are extremely different, for example, when an image captured by an imaging device is to be used under a printing plate, the pixel density of an image obtained by imaging by the imaging device is one screen. The pixel density of an image on an electronic plate making machine is (thousands × thousands) per screen, while the maximum is about (500 × 500) per screen.
As described above, since the size of the image obtained by the above-described image capturing apparatus is orders of magnitude larger than that of the image obtained by the image capturing apparatus, even if the compression / expansion method of the image information corresponding to the pixel is not performed at all, an alias is generated by the pixel expansion. I do.

If interpolation is performed without performing the above-described pixel enlargement, a large interpolation area is applied with a weighted average value of known data. Image quality degradation cannot be avoided. Conversely, when the pixel density of the original image is (thousands × thousands) per screen, the correlation between adjacent pixels is extremely high. Although compression is possible, the compression ratio cannot be increased by the conventional image information compression method that requires that the number of pixels between the original image and the restored image (expanded image) match as described above. Disadvantage occurs.

[0007] In order to solve the above-mentioned problems, the applicant of the present invention disclosed in Japanese Patent Application No. 5-39492, regardless of the pixel density of an image to be subjected to image information processing. Only the characteristic points of the image are extracted to obtain compressed image data of the image information, and when decompression, pixel restoration is not performed from the above-described image data. The luminance function of the above-described image information is obtained for two-dimensionally distributed luminance information and three-dimensionally distributed image information including a time axis in the two-dimensionally distributed luminance information so that the image can be drawn. The characteristic points are the positive and negative maximum points of the curvature of the isoluminance line, or the positive and negative maximum points of the curvature of the isoluminance surface, or a straight line obtained by linearly approximating the luminance function isoluminance line and the luminance function isoluminance line described above. Error or brightness function An error between a plane obtained by planar approximation of the luminance plane and the above-described luminance function equi-luminance plane is determined as a feature point of the image at a point exceeding a predetermined threshold, and the positions of the characteristic points of the image and the luminance value are determined. When transmitting and recording, and using the positions of the characteristic points and the luminance values for image restoration, the luminance information of the pixels other than the characteristic points is extracted by an interpolation plane or an interpolation solid determined by a plurality of neighboring characteristic points upon decompression. The image information of the luminance information that is two-dimensionally distributed is determined from the three-dimensionally distributed image information including the time axis in the image information of the two-dimensionally distributed luminance information.
Targeting a plurality of sets of luminance information distributed in a three-dimensional manner, a straight line obtained by linearly approximating the positive / negative maximum value of the curvature of the equiluminance line of the luminance function of each of the sets, or a linear approximation of the isoluminance line of the luminance function of each of the sets. And a point at which an error between the above-described luminance function and the isoluminance line exceeds a predetermined threshold value is set as a feature point of the image, and the position and the brightness value of the feature point of the image are transmitted, recorded, and restored. When used, a multidimensional image compression / expansion method is provided in which the luminance information of pixels other than the characteristic points is determined by an interpolation solid determined by a plurality of nearby characteristic points upon expansion.

The following is a description of the multi-dimensional image compression / expansion method already proposed. Now, assuming that the luminance in the monochrome still image is z, the horizontal direction of the screen is x, and the vertical direction is y, the image can be generally expressed by the following equation (1). z = f (x, y) (1) In addition, if the time axis is t, the moving image can be represented by the following equation (2). z = f
(X, y, t) (2) Here, if f is a multi-dimensional function, the luminance z in the image is represented by the following equation (3).

[0009]

(Equation 1)

[0010] By the way, transmitting an image means
It can be said that the above-mentioned luminance function determined on the image transmitting side is reproduced on the image receiving side.However, in general digital image transmission, the luminance function is not analyzed analytically, and it is used as a so-called table function. All of the table values are transmitted. On the other hand, in the conventional general compressed transmission, high-efficiency coding is performed using the adjacent correlation of the table value itself, or the same measure is applied to the table value after the orthogonal transformation. Although measures such as rubbing are taken, there are few conventional compression transmission systems which directly extract characteristic values of functions from analytical processing relating to luminance functions.

Image information such as two-dimensional luminance information in a still image or three-dimensional luminance information including a time axis in the two-dimensional luminance information is exemplified in FIG. 26C. A luminance line (contour line) {an isoluminance plane in the case of three-dimensional luminance information} is extracted, and the extracted isoluminance line is extracted.
(Contour line) {In the case of three-dimensional luminance information, the positive or negative maximum point of the curvature of the isoluminance plane｝ or the above-mentioned contour line (contour line)
Linear approximation of {equi-luminance plane for three-dimensional luminance information} (3
In the case of two-dimensional luminance information, the error from the plane approximation｝
A point exceeding a predetermined threshold value is used as a feature point of the image, and the position and the brightness value of the feature point of the image are transmitted (recorded) and used for image restoration. By rejecting the luminance information of a small number of pixels, it is possible to realize a large compression of image information, and at the time of expanding image information, an interpolation plane determined by a plurality of nearby feature points transmitted (recorded) is used. , Or high-quality image decompression by interpolation solid.

That is, in the proposed multidimensional image compression / expansion method, first, when the image information is compressed with respect to a still image, the above-mentioned feature points extracted from the image, that is, the contour lines (contour lines) of the luminance function. ) {A three-dimensional luminance information, in the case of three-dimensional luminance information, a positive / negative maximum point of the curvature of the equal luminance plane}, or the above-mentioned equiluminance line (contour line) {in the case of three-dimensional luminance information, a linear approximation of the equal luminance plane} (In the case of three-dimensional luminance information, a feature point of an image when an error from the plane approximation｝ also extracts a point exceeding a predetermined threshold,
1. When the curve of the above-described contour line {in the case of three-dimensional luminance information} before and after the pixel a to be considered is more than a predetermined threshold angle,
The pixel a is determined to be a feature point. 2. According to a virtual straight line between the already detected feature point pixel and the contour line (the contour plane in the case of three-dimensional luminance information) therefrom, the pixel is traced in a certain direction. Among the pixels between the two pixels, a pixel indicating a distance exceeding a certain threshold distance is determined to be a feature point.

FIG. 26 illustrates a luminance distribution function in a two-dimensional screen of an arbitrary still image using the above-described determination criterion. A specific example of the case of extracting points is as follows. FIG. 26A illustrates one of the contour lines of the luminance distribution function in the two-dimensional screen. The extraction of the contour lines of the luminance distribution function in the two-dimensional screen is as follows. It is done by a procedure. FIG.
With reference to (b), a method of searching for a point having the same luminance will be described. In FIG. 26B, (1) to (4) are four pixels, and the luminance values of the pixels (1) to (4) are H and L (H is a high luminance value, L is a low luminance value), and the point S at which the first extracted value is obtained is set as a contour line extraction start point. Said S
Since the point is located between the pixel (1) and the pixel (4), the luminance value at the coordinate position is, for example, a luminance value corresponding to the luminance value of the pixel (1) and the luminance value of the pixel (4). A proportional distribution value may be used.

An isoluminance line (contour line) obtained by connecting a point indicating a luminance value equal to the luminance value of the point S from the start point S described above.
In order to track, the sides (1)-(2), (2)-in the image shown by the four pixels (1) to (4) described above.
When it is verified in the order of (3), (3)-(4) whether or not there is a point having the same luminance value (contour value) as the luminance value of the above-mentioned start point S, the above-mentioned contour value is determined by the side (1)- (2) It is detected at the upper M1 point. After the point M1 having the above-mentioned contour value is detected, the above procedure is repeated to detect the next contour point M2, so that the points showing the contour values are sequentially traced to extract the contour line. Go ahead. In the extraction of the contour lines, the sides where the contour lines already intersect are excluded from the detection targets (only the above-mentioned start point S is an exception). As described above, by detecting successive points having a certain luminance value and an equal luminance value, one of the contour lines of the luminance distribution function in the two-dimensional screen as illustrated in FIG.
One is extracted.

[0015] One of the contour lines of the luminance distribution function in the two-dimensional screen
In FIG. 26A exemplifying one of the points, a point where the course of an isoluminance line (contour line) formed by a series of isoluminance points sequentially tracked from the start point S is rapidly changing. M6 is the feature point criterion 1. (Points M8, M14, M16, M18,
For M20, the criterion for determining the feature point described above is as follows. Is determined as a feature point according to After the point M8 indicating the above-mentioned equal luminance value is detected, when the points of the equal luminance value are sequentially tracked and the point M12 is reached, the point M10 at which the distance is generated with respect to the virtual straight line M8-M12 is in the direction. Although it is not a sudden change point of the above, the criterion for the specific point described above2. Is determined as a feature point according to This is the same for the points M12 and M12.
As described above, since the luminance distribution function in the two-dimensional screen is replaced with the contour lines (contour lines), the image information is rejected and further concentrated on the characteristic points of the contour lines. In the image compression / expansion method, transmission (recording) can be performed in a highly compressed state.

As described above, in the proposed multidimensional image compression / expansion method, the two-dimensional luminance function in the image to be compressed is replaced with the position information and luminance information of a small number of feature points. Thus, in order to expand (restore) the image information transmitted (recorded) in a state where the image information is highly compressed and obtain a reproduced image, the above-described feature points correspond to the corresponding points ( (It is not necessary to correspond to the original image and pixels)
Is determined, and the decompression method is adopted such that the influence on peripheral pixels is lost in proportion to the distance from the feature point. The above-described expansion of image information is performed by linear interpolation between feature points on one dimension, but expansion of image information in a multidimensional space is performed by plane interpolation or stereoscopic interpolation. Next, the principle will be described with reference to FIG.

FIG. 27 shows characteristic points S, M6, M8, M10, M12, M14, and M16 shown in FIG. 26A obtained as described above with reference to FIG. , M18, M2
Contour lines (constant luminance lines) reproduced on the reproduction side image memory using the position information (address) of 0 and M22 are included. In FIG. 27, the contour lines (the contour lines of the M group) are the characteristic points S, M6, M8, M10, M12, M14, M16,
M18, M20, and M22 are indicated by dotted lines that sequentially connect them. In FIG. 27, each feature point Ni, Nj, Nk, Nl
Are the contour lines of the feature points of the N group different from the contour lines of the feature points of the M group described above.

Now, when the area between the contour lines of the M group and the contour lines of the N group is filled with the luminance value indicated by the N group, the luminance value in the area surrounded by the contour lines of the M group is obtained. A so-called false contour phenomenon occurs due to a clear luminance step occurring between the images, so that a high-quality expanded image cannot be obtained. Therefore, in order to prevent the above-described pseudo contour phenomenon from occurring, the feature points belonging to the M group and the feature points belonging to the N group are represented by za = Σ (zi
/ Ri) / Σ (1 / ri) (A) Applicable by plane interpolation, which is an approximation of the above equation (A).

Here, considering a specific pixel a in the expanded screen,
The distance from the specific pixel a to the pixel of each feature point is r
i, the luminance value of the pixel at each feature point is zi, and α is a proportionality constant. Is represented by the following interpolation formula (4). za = zk + αk · rk (4) αk · rk in the second term on the right side in the above equation (4) indicates a difference between the luminance value of pixel a and the luminance value zk of feature point k, The value of αk · rk is proportional to the distance r. The luminance value of the pixel a and the luminance value zk of the feature point k
The difference αk · rk between the luminance value and the positive and negative values is reflected in the proportionality constant α, but the sum of α in the entire interpolation space is zero. Therefore, Σαk = 0 (5)
Eliminating the term of α from the above equation (4) in consideration of the condition of the equation (5), a general interpolation equation for the luminance value za of the pixel a is obtained, and the above equation (A) is obtained. It is done. za = Σ (zi / ri) / Σ (1 / ri) (A)

The above equation (A) is used to calculate the luminance values zi of all the feature points.
And the distance r from the pixel a to be interpolated is known,
This indicates that the luminance values of all uninterpolated pixels inside the screen (there is no need to correspond to the original image) can be obtained by interpolation. However, since the expression (A) exceeds the practical amount of calculation depending on the increase of the feature points, the interpolation luminance value is approximately calculated using an interpolation plane determined by three neighboring feature points surrounding the uninterpolated pixel. Since one plane is determined by three points in space, as is well known, an interpolation triangle can be obtained by grouping three nearby feature points. Therefore, three feature points “Ni, Nj, S”, “Nk, M6, S”, “Nj, Nk,
The luminance value of the pixel in each triangle formed by "S"
It can be applied by the plane interpolation by the approximation operation performed by the equation (A).

In order to determine the interpolation plane when performing the above-described plane interpolation, it is necessary to extract three nearby feature points. Next, referring to FIG. The feature point extraction principle will be described. In FIG. 29, pixels Mi and Mj are M pixels whose luminance value (equal value) is M.
Pixels Ni, Nj, and Nk are characteristic points on the contour line (contour line) of group N whose luminance value (contour value) is N. And the pixels Oi, 0j, and 0k are characteristic points on the contour line (contour line) of the O group whose luminance value (contour value) is O, and have the above-mentioned respective contour values when the image is expanded. The data of the feature point group is stored in the reproduction-side image memory. In the figure,
Each of the above-mentioned contour lines is indicated by a separate dotted line.

In order to search for an arbitrary pixel other than a feature point as to which feature point the arbitrary pixel is closer to,
For example, it can be performed as shown in FIG. FIG.
In the case where it is assumed that the pixels M and N of the characteristic point exist in the vicinity of the arbitrary pixel P, the adjacent pixels are spirally inspected from the arbitrary pixel P according to the path indicated by the dotted arrow in the figure. Reach the pixel M of the feature point before the pixel N of the feature point. As a result, the above-mentioned arbitrary pixel P is detected not as the pixel N of the feature point but as being in the vicinity of the pixel M of the feature point. Therefore, a mark indicating that the pixel P is located in the region of the pixel M of the feature point is entered.

FIG. 29 shows a state in which a plurality of arbitrary pixels other than the pixel of the characteristic point are provided with a characteristic point symbol to which each arbitrary pixel belongs according to the procedure described above with reference to FIG. Have been. The dotted line connecting the above-mentioned feature point symbols is the boundary of the influence area of the above-mentioned feature point. now,
For example, when attention is paid to Δ1, which is a boundary between three regions, this Δ1
Are adjacent to pixels Nj, Nk, Ok. Therefore, the pixels Nj, Nk, Ok are regarded as three nearby feature points,
The triangles Nj, Nk, and Ok are determined as the luminance interpolation plane, and the luminance values of the pixels inside the triangles Nj, Nk, and Ok are used by the luminance values of the luminance interpolation plane. Similarly, Δ2 which is the boundary of another three areas
, Pixels Nj, Oj, Ok are adjacent to Δ2. Therefore, the pixels Nj, Oj, Ok are located near
, And the triangles Nj, Oj, Ok are determined as a luminance interpolation plane, and the triangles Nj, Oj,
The luminance value of the pixel existing inside Ok is filled with the luminance value of the luminance interpolation plane described above.

According to the above-described procedure, the pixel area affected by one feature point is searched for on the extended plane (extended solid in the case of a three-dimensional object), and the pixel area affected by the one feature point is searched. By detecting three neighboring feature points (four feature points in the case of a three-dimensional object) from the boundary, a luminance function on the expansion side is determined from a small number of characteristic point information, and each luminance of untransmitted image information is determined. The values are applied by the interpolation values to reproduce the contents of the entire image. In addition, a circle, a square, a rectangle, a rhombus, or the like may be used as an extended surface in the case of a two-dimensional object, and a sphere, a spheroid, and a base are symmetrical in an extended solid in the case of a three-dimensional object. Two cones connected as surfaces, two pyramids connected with a bottom surface as a symmetric surface, or the like may be used.

According to the above-described procedure, the luminance function on the expansion side is determined based on the small number of feature point information, and each luminance value in the untransmitted pixel information is interpolated by the luminance value on the luminance interpolation plane. Thus, the contents of the entire image are reproduced. As is clear from the above description, what is to be shared between the compression side and the decompression side is the brightness function of the screen. The number of pixels on the side does not need to match. That is, the feature information of the image extracted by the proposed multidimensional image compression / expansion method is redrawn according to the number of pixels (resolution scale) on the expansion side.

FIG. 25 is a diagram showing an example of the configuration of a contour tracer 7 that can be used in the proposed multidimensional image compression / expansion method. The image memory 27 in FIG. 25 stores a predetermined number of pixels (for example, 512 pixels in the horizontal direction of the image, and A digital signal having a predetermined number of bits (for example, 8 bits) is stored for each pixel in a state of being decomposed into 480 pixels in the direction. The above image memory 2
Reference numeral 7 denotes an image memory used to extract contour lines of a specific luminance in an image and perform an operation for determining a feature point. The output from the image memory 27 is supplied to the contour tracer 7. The above contour tracer 7
Then, extraction of contour lines (contour lines) in image information and determination of feature points are performed.

The contour line tracer 7 shown in FIG. 25 has a plurality of (n) address generators 28 to 30, luminance determiners 31 to 33, luminance value generators 34 to 36, shift registers 37 to 39, and features. Point determiners 40 to 42 and one alignment circuit 43 are provided. The address generators 28 to 30 generate address data for specifying an address of the image memory 27 and provide the generated address data to the image memory 27.

The luminance value data read out from the storage area of the address specified by the address data generated by the address generator 28 (29,... 30) and the luminance value data generated by the luminance value generating section 34 (35,. The luminance determination unit 31 (32,... 3) to which the data of the specified specific luminance value
In 3), the two data are compared with each other, and the comparison result is supplied to the address generators (28,..., 30). Thereby, the address generator 28 generates new address (including the middle of the pixel address) data corresponding to M1 in FIG. 26B, for example (the other address generators 29,..., 3).
It goes without saying that address data is generated even if it is 0), and is given to the image memory 27. In the image memory 27, the data of the luminance value read from the storage area of the address specified by the address data given as described above is supplied to the luminance determiners 31 (32,... 33) of the contour tracer 7. .

The luminance judging units 31 (32,... 33) read the luminance value data read from the image memory 27 and the specific luminance value data generated by the luminance value generating unit 34 (35,. , And the comparison result is supplied to the address generators (28,..., 30). Thereafter, the above-described operations are sequentially performed in the same manner, whereby the luminance value substantially equal to the luminance value indicated by the specific luminance value data generated by the luminance value generating unit 34 (35,..., 36) is obtained. Are sequentially generated. When each of the above address values indicates the middle of the pixel address value as in the example illustrated in FIG. 26, the above operation is sequentially performed by replacing the pixel address value with a neighboring pixel address value. .

As described above, the address generator 28 (29,
.. 30) are also given to the shift registers 37 (38,... 39), and the parallel outputs of the shift registers 37 (38,. In the determiners 40 (41,..., 42), one new address is input in a state where the shift register 37 (38,..., 39) stores address data of a specific number of pixels (for example, 10 pixels). Each time one of the oldest addresses disappears, a pixel serving as a feature point is detected from the address group in accordance with the above-described feature point determination criteria.
Then, the address value and the luminance value of the pixel serving as the feature point are supplied to the alignment circuit 43. In the alignment circuit 43, as described above, the address value and the luminance value of the feature point group output from the plurality of characteristic point determiners 40, 41,... An alignment operation for sorting and sending is performed. The output from the alignment circuit 43 is
It is sent as the output of the contour tracer 7. Note that the image memory 27 has a multi-port structure because the signal processing in the contour tracer 7 tracks a plurality of contours in parallel. In terms of time, when scanning for one screen (one field period for an image employing the interlaced scanning method as a scanning standard) by the contour tracer 7 is completed, the characteristic points for the one screen are determined. The extraction operation ends.

By the way, the above-mentioned equiluminance line of the luminance information,
Alternatively, when obtaining an equi-luminance plane of luminance information, it is necessary to track a pixel having a specific luminance value, but the center of a boundary pixel is tracked as in a conventional general image contour tracking method. In such a case, an error occurs in the outline between the original image and the reproduced image, the processing time increases due to an increase in the processing amount, and the center of the boundary pixel is tracked. When a chain code code string is used, when a straight line having a width of one pixel protrudes, the center of one boundary pixel is caused to correspond twice. Since the straight line has a problem that the pixel width is one pixel, the present applicant has proposed a binary image contour tracking method as disclosed in Japanese Patent Application Laid-Open No. 5-35872. did.

In the contour tracing method for a binary image disclosed in the above-mentioned Japanese Patent Application Laid-Open No. 5-35872, the tracing direction of the boundary point is defined by white / black of four pixels around the boundary point of the pixel. The tracking direction is determined such that there is a black pixel on the right side and a white pixel on the left side. When the tracking direction is upward, "1"; when left, "2"; when downward, "3"; At this time, the tracking direction in each case is assigned to each bit of the 4-bit tracking direction flag, and a bit having a tracking direction is set to "1", and a bit having no tracking direction is set to "0". Attaching the tracking direction flag to all the boundary points, searching the boundary points from the tracking direction flags, and tracing along the tracking direction from the starting point as a starting point, and tracking the tracked boundary points Change direction flag to line , All of the tracking direction flag is "0
000 "to obtain the outline of the image.

In the above-described binary image contour tracking method, a boundary pixel (boundary point) can be tracked only by applying a predetermined tracking direction to an input image, so that the tracking direction is searched each time. There is no need for a large-capacity memory and the processing time is short, but the tracking direction is determined so that there are black pixels on the right side and white pixels on the left side in the tracking direction, and the contour tracking follows the pixel boundary line Therefore, when the reproduced image is viewed, a black pixel of one pixel is generated on the left side in the tracking direction. The above point does not matter at all when the original image is, for example, a character, but when the tracking target is a general image, the image quality may be degraded by the above point, for example, In the case where the binary image contour tracking method is applied to the already proposed multidimensional image compression / expansion method described above, a solution is required.

Further, in the multidimensional image compression / expansion method already proposed, the isoluminance line (or isoluminance surface) is approximated by a polygon (or polygon), and the polygon is approximated by a straight line connecting the characteristic points at the time of expansion. (Or polygonal approximation)) When luminance information of a pixel other than a feature point is determined by an interpolation plane (or an interpolation solid) determined by a plurality of nearby feature points on the equalized luminance line (or an equal luminance plane) And the above-mentioned polygon-approximate isoluminance line (or isoluminance plane)
Is different from the original isoluminance line (or isoluminance plane) due to the compression of the amount of information.
In some cases, the order of the luminance order is reversed between the images, and an abnormal state occurs in the luminance distribution in the reproduced image, resulting in deterioration of the image quality. Therefore, a solution has been sought.

[0035]

SUMMARY OF THE INVENTION According to the present invention, a feature point which meets a specific condition with respect to an isoluminance line set for each predetermined specific luminance value in a luminance function of an image is obtained, and the characteristic of the image described above is obtained. In the multi-dimensional multi-valued image compression / expansion method used for transmission, recording, and image restoration of point positions and luminance values, a specific luminance boundary is obtained by tracing the contour of a pixel with a specific luminance value during compression. At the center of each pixel
In the multidimensional multi-valued image compression / expansion method in which the coordinates of the specific isoluminance line passing point are moved, and in the above-described multidimensional multivalued image compression / expansion method, in order to obtain an equal luminance line for each specific luminance value upon decompression. For each of the threshold values, a bright area bordered by the isoluminance line is filled with a bright symbol, and a dark area bordered by the isoluminance line is painted with a dark symbol to create a mask by a bright / dark binary symbol. In the multidimensional image compression / expansion method, and the multidimensional image compression / expansion method described above, at the time of decompression, for each threshold value for obtaining an equal luminance line for each specific luminance value, the equiluminance line is used as a boundary. A bright area is painted with a bright symbol, and a dark area bordered by the isoluminance line is painted with a dark symbol to create a mask with a bright / dark binary symbol. Arrangement And, the region between the adjacent mask,
After filling with an intermediate value of the luminance values based on the threshold values of the two adjacent masks described above, the relationship between the luminance value of each pixel and the luminance values of eight surrounding pixels having the respective pixels as the central pixels is represented by (1) When the brightness value of the surrounding pixels is equal to the brightness value of the center pixel, the brightness value of the center pixel is undecided.
(2) When the luminance value of the surrounding pixels is not higher than the luminance value of the central pixel, the luminance value of the central pixel is set to “the intermediate value of luminance values based on the threshold values of both adjacent masks−a” (where a is a positive value). (3) When the luminance value of the surrounding pixels is not lower than the luminance value of the central pixel, the luminance value of the central pixel is set to “the intermediate value of luminance values by threshold values of both adjacent masks + a”. (4) When the relationship between the luminance value of the surrounding pixels and the luminance value of the central pixel is other than the above-described relationships (1) to (3),
An intermediate value between the luminance values based on the threshold values of both adjacent masks,
It is determined according to which of the conditions shown in the above (1) to (4) is satisfied, and the area center line is extracted for the undetermined luminance value area, and the pixel on the area center line described above is extracted. A multidimensional multi-valued image compression / expansion method in which the luminance value is an intermediate value of the luminance values by the threshold values of both adjacent masks, and for the luminance value undetermined region as described above, the region center line is extracted, and the region center line is extracted. The luminance value of a pixel whose luminance value is undetermined after the luminance value of the pixel is set to an intermediate value between the luminance values based on the threshold values of the two adjacent masks is the known luminance at both ends of the extension line in the fixed direction of the pixel whose luminance value is undetermined. The luminance value is determined by the primary interpolation method using the luminance value and the distance of the pixel of
Alternatively, the luminance value of the pixel whose luminance value is undetermined is first-order-interpolated using the luminance value and the distance of the pixel of known luminance at both ends of the straight line in each fixed direction in each different direction of the pixel whose luminance value is not determined. The surface interpolation value determined by the average value of the luminance values obtained by the method, or the luminance value of the remaining luminance value undetermined pixels is obtained by using three pixels whose luminance values near the luminance value undetermined pixels are known. And a multi-dimensional multi-valued image compression / expansion method determined by the following.

[0036]

The compression / expansion of a multi-valued still image will now be described, taking as an example the case where the isoluminance plane showing a specific luminance value in the luminance function of the original image to be compressed is represented by FIG. The case will be described as an example. In FIG. 20, pixels surrounded by a dotted line are white pixels having a luminance value higher than the luminance value of the equi-luminance surface, and pixels surrounded by a solid line and shaded It is assumed that the pixel is a black pixel having a lower luminance value than the luminance value. With respect to the pixel group shown in FIG. 20 described above, the boundary line between the white pixel and the black pixel is formed by applying the outline tracking method of the binary image disclosed in the above-mentioned Japanese Patent Application Laid-Open No. 5-35872. Upon tracking, the outline of the image as shown in FIG. 21 is obtained.

That is, the tracking of the boundary line is firstly performed by
When scanning is started from the upper left corner of the screen according to the raster scanning order and scanning is performed, the first boundary designated as the first boundary in FIG. 20 is found. When the tracking is started with the above point as the starting point, if the high-luminance surface is located on the left side of the tracking direction, if the tracking start direction matches the horizontal scanning direction (CW: clockwise), the tracking starts. The direction coincides with the vertical scanning direction (CCW: counterclockwise). In FIG. 21, tracking direction type flags (CW, CCW) are provided for each tracking equal luminance line. As illustrated in FIG. 21 described above, after tracing and obtaining the contour of a pixel for each predetermined specific luminance value in the luminance function of the image, the specific luminance boundary is determined as illustrated in FIG. Are moved to the center position of each pixel constituting the line.

That is, when the tracking start direction coincides with the vertical scanning direction (CCW: counterclockwise),
It is shifted to the pixel center position to the left in the tracking direction, and the isoluminance line when the tracking start direction matches the horizontal scanning direction (CW: clockwise) is shifted to the pixel center position to the right in the tracking direction. As described above, when a reproduced image is obtained based on a specific luminance boundary obtained by tracking the contour of a pixel,
A black pixel of one pixel is generated on a specific side with respect to the tracking direction, thereby deteriorating the image quality. However, as described above, each of the specific luminance boundaries obtained by tracking the contour of the pixel for each specific luminance value By moving the coordinates of the passing point of the specific isoluminance line to the center position of the pixel, the improvement can be made satisfactorily.

Next, the starting point is fixed, and the pixels on the isoluminance line are successively connected by a virtual straight line from the starting point, and the distance between the virtual straight line and a real pixel existing therebetween is predetermined. An actual pixel exceeding the error allowable range E is registered as a feature point pixel. When a feature point pixel is found as described above, the feature point pixel is set as a new starting point, pixels on the isoluminance line are sequentially connected from the new starting point by a virtual straight line, and the virtual straight line is A real pixel whose distance to a real pixel existing between the real pixels exceeds a predetermined error allowable range E is registered as a feature point pixel. Hereinafter, successive feature points are sequentially found and registered using successive feature point pixels as successive start points. A new contour line obtained by connecting the characteristic points sequentially obtained as described above is as illustrated in FIG.

By performing the above operation for each of the equal luminance planes having different luminance values, the luminance function of the original image is reduced to the data of the characteristic points of the equal luminance line. Compression will be performed. The position and luminance value of the feature point of the image are used for transmission, recording, image restoration, and the like of the original image information as highly efficient compressed information. On the extension side, the equal luminance line can be easily drawn and restored on the extension side. The important point here is that since the isoluminance line restored on the decompression side is drawn by the end point (feature point) designation vector, the pixel density of the image memory provided on the decompression side is set on the compression side. Even if the pixel densities of the image memories used are different, the shape of the isoluminance line restored on the expansion side is hardly influenced by the physical conditions as described above.

As described above, the information obtained by decompressing the information efficiently compressed to the information of the position of the characteristic point and the luminance value of the image meeting the specific conditions obtained for each specific isoluminance line is obtained to obtain a reproduced image. At this time, for each threshold for obtaining an equal luminance line for each specific luminance value,
A bright area with the above-mentioned isoluminance line as a boundary is filled with a bright symbol, and a dark area with the above-mentioned isoluminance line as a boundary is painted with a dark symbol to create a mask with a binary symbol of light and dark. FIG. 24 shows a mask created by painting out the pixels on the decompression side through which the equal luminance line at a specific luminance value has passed, and the area on the right side (low luminance side) in the tracking direction.

As described above, the brightness 2
By creating a mask with a value symbol, the isoluminance line (or isoluminance surface) is approximated by a polygon (or polygon),
Equiluminance line (or equiluminance surface) approximated by a polygon (or polygonal approximation) by a straight line connecting feature points during expansion
In the case where the luminance information of pixels other than the characteristic points is determined by the interpolation plane (or the interpolation solid) determined by the plurality of nearby characteristic points, the equiluminance line (or the luminance plane) is used.
No inversion of the brightness order between
It does not cause an abnormal state in the luminance distribution in the image to cause deterioration in image quality.

FIG. 12 shows a case where the gradation of the luminance of the image is 256 (000 to 255) by an 8-bit digital luminance signal, as shown in FIG. An example is shown in which nine types of masks for converting the above-described 256 levels of luminance into nine types of luminance levels are formed by the first to eighth threshold values having the appropriate luminance values as threshold values. FIG. 6 shows an example in which a mask having a mask number corresponding to the mask number shown in FIG. 12B is created for an original image having a luminance gradation of 256. . In FIG. 6, the screen displayed as Mask 1 is entirely white, but as shown in FIG.
The luminance value corresponding to 30 in the 6 gradations is set, and a portion where the luminance value of the image is higher than the threshold is painted white, and a portion where the luminance value of the image is lower than the threshold is painted black. Since the mask in the compressed state is created, the fact that the entire screen displayed as mask 1 in FIG. 6 is shown in white means that the image to be compressed has 256 gradations. This means that there is no portion lower than the luminance value corresponding to 30.

Further, the screen displayed as the mask 8 in FIG. 6 is entirely black, but FIG.
As shown in the figure, the mask 8 has a threshold value set to a luminance value corresponding to 240 out of 256 gradations, a portion where the luminance value of the image is higher than the threshold value is painted out in white, and Since a portion in which the luminance value of the image is lower than the threshold value is created in a state where the mask is painted black, FIG.
The fact that the entire screen displayed as the middle mask 8 is shown in black means that the image to be compressed has a portion higher than the luminance value corresponding to 240 out of 256 gradations. Means not. Further, the screens shown as masks 2 to 7 in FIG. 6 are partially black in the drawing, which is more than the threshold values used to create masks 2 to 7. This means that the high luminance value and the low luminance value are mixed in the image to be compressed.

When a plurality of masks indicated by the respective mask numbers shown in FIG. 12B and FIG. 6 are superimposed, an image having nine levels of luminance is reproduced. Now, in each mask indicated by each mask number shown in FIG. 12 (b) and FIG. 6, the luminance value of the portion painted white is expressed by the numerical value of each mask number. FIG. 7 shows the luminance distribution of an image obtained by superimposing a plurality of masks indicated by mask numbers 1 to 8 in FIG. 6 by numerical values.

In FIG. 12B, the luminance of 256 gradations indicated by the 8-bit digital luminance signal is compressed into luminance values corresponding to the respective thresholds defined by mask numbers 0 to 8. It is in a state of being left. In general, assuming that the mask group is n masks, mask number 0 → pixel of luminance value 0 to first equal luminance value Mask number 1 → first equal luminance value to second equal luminance value Pixel of mask number 2 → second equal luminance value to third equal luminance value Pixel of mask number 3 → third equal luminance value to fourth equal luminance value Pixel:::::: Mask number i → second Pixels with i equal luminance value to (i + 1) th equal luminance value:::::: mask number n → nth equal luminance value to 255th pixel That is, if n masks are created as described above, the luminance values of each pixel distributed from luminance 0 to luminance 255 are classified into n + 1 categories. The twelfth (b) example described above is an example where n = 8.

When the luminance of the pixels included in each of the above categories is represented by an intermediate representative value of the luminance range, the twelfth level (b) and the example of FIG. Will be done. Next, from among the plurality of isoluminance lines, three neighboring pixels are selected from two adjacent luminance lines, and the above-mentioned 2 pixels are selected on the interpolation plane formed by the three pixels.
When the luminance between two equal luminance lines is interpolated, a state as shown in FIG. 11A is obtained by one-dimensional display. Now, although the original image has 256 gradations, for example, in the above-described example, the image is expanded by interpolation in a state where the luminance gradation is reduced to 9 luminance levels. The quality of the reproduced image is not sufficient.

Therefore, in order to improve the quality of the reproduced image, in the interpolation reproduction of the luminance value, the luminance value of the interpolation end point is not set to the intermediate value of the simple category as described above. After arranging the masks in the order of the magnitudes of the luminance values according to the threshold values and filling the area between the adjacent masks with the intermediate value of the luminance values according to the threshold values of the two adjacent masks, the luminance value of each pixel is (1) When the luminance value of the surrounding pixels is equal to the luminance value of the central pixel, the luminance value of the central pixel is undecided. , (2) when the luminance value of the surrounding pixels is not higher than the luminance value of the central pixel,
The luminance value of the center pixel is defined as “an intermediate value of luminance values based on threshold values of both adjacent masks−a” (where a is a positive value).
(3) When the luminance value of the peripheral pixel is not lower than the luminance value of the central pixel, the luminance value of the central pixel is set to “the intermediate value of luminance values by threshold values of both adjacent masks + a”. When the relationship between the luminance value of the pixel and the luminance value of the center pixel is other than the above-described relationships (1) to (3), the luminance value is set to an intermediate value between the luminance values based on the threshold values of both adjacent masks. The determination is made according to which of the conditions shown in (1) to (4) is satisfied, so that the quality of the reproduced image can be improved.

A specific example in which the relationship between the center pixel and the luminance values of the eight surrounding pixels is (1) to (4) described above is shown by surrounding four dotted lines shown in FIG. Each of the four dotted line frames shown in FIG. 7 is taken out for easy understanding and shown in FIG. By the way, when the relationship between the central pixel and the luminance values of the eight surrounding pixels is (1),
That is, when the luminance value of the surrounding pixels is equal to the luminance value of the central pixel, the luminance value of the central pixel is undecided.
The luminance value of the nine pixels is the same numerical value 4. In this case, the luminance value of the center pixel is “undefined”, and in FIG. 9 and subsequent figures, “undefined” is indicated by a symbol “•” for simplification of the drawing.

When the relationship between the central pixel and the luminance values of the eight surrounding pixels is the above (2), that is, when the luminance value of the peripheral pixels is not higher than the luminance value of the central pixel, the central pixel Is set to “the intermediate value of luminance values based on the threshold values of both adjacent masks−a” (where a is a positive value), the dotted line shown in FIG. Although shown by nine pixels in the frame, the luminance value of the central pixel in such a case is assumed to be “low”, and in FIGS. Indicates the state of "low" by using the symbol of "-".

Further, when the relationship between the central pixel and the luminance values of the eight surrounding pixels is the above (3), that is, when the luminance value of the peripheral pixels is not lower than the luminance value of the central pixel, FIG. 8 shows an example in which the luminance value of a pixel is set to “the intermediate value of luminance values by threshold values of both adjacent masks + a”
, The luminance value of the center pixel in such a case is assumed to be “high”, and is shown in FIG. 9 and subsequent figures. For the sake of simplicity, the state where the luminance of the center pixel is “high” is indicated by a “+” symbol.

Further, the relationship between the central pixel and the luminance values of the eight surrounding pixels is the case of the above (4), that is, the relationship between the peripheral pixel luminance value and the central pixel luminance value is as described above. In the case other than the relations (1) to (3), an intermediate value between the luminance values based on the threshold values of both adjacent masks is used. In this case, the luminance value of the central pixel is “intermediate”, and in FIG. 9 and subsequent figures, the luminance of the central pixel is “intermediate” in order to simplify the illustration.
Is indicated by using a symbol “=”. FIG. 9 illustrates the relationship between the luminance values of the respective pixels shown in FIG. 7 using the symbols “·”, “−”, “+”, and “=”. a) to (l) show the brightness values in a three-dimensional manner in the height direction so that the relationship between the center pixel and the brightness values of the eight pixels around the center pixel and how to use the above-mentioned symbols can be clearly understood. FIG.

Next, with respect to the indeterminate brightness value area, an area center line (skeleton), that is, a group of points equidistant from the edge of the area is extracted, and the brightness values of the pixels on the area center line are compared with the adjacent masks. Is an intermediate value of the luminance values based on the threshold value.
FIG. 13 to FIG. 16 are diagrams illustrating the above process. FIG. 13 shows the luminance values before correction only of the indefinite area indicated by using the symbol “•” in FIG. 9, and FIG. FIG. 15 is a diagram showing a state in which an intermediate mark = and an intermediate luminance value are added to pixels on the center line of the indefinite area, and FIG. 16 shows all the marks attached in accordance with the above-described rules. FIG. 6 is a diagram showing the luminance value of each pixel when used.

FIG. 12A shows the relationship between the luminance values when the luminance value of the pixel is “intermediate level”, “intermediate luminance−”, and “intermediate luminance +”. In FIG. 12A, the difference between the threshold values of the adjacent masks described above, that is, the difference between the i-th threshold value and the (i + 1) -th threshold value is indicated as d. The intermediate level shown in FIG. 12A is an intermediate luminance value between the luminance values defined by the threshold values of the adjacent masks described above. The luminance level of the low level “intermediate luminance−” shown in FIG. 12 is set to a luminance level lower by d / 3 from the above-mentioned intermediate level, and furthermore, the high level shown in FIG. The luminance level of the level “intermediate luminance +” is a luminance level higher by d / 3 than the above-mentioned intermediate level. And FIG.
To the right of 2 (b), for each mask number,
Intermediate level, low level, high level values, display symbols and the like are shown.

FIG. 17 shows the luminance value of each pixel using the intermediate level, low level, and high level values for each mask number shown in FIG. 12B. In FIG. 17, the luminance value of the pixel is still “undefined”, and there is a pixel indicated by the undefined mark “•”. As described above, after the luminance value of the pixel on the region center line is set to an intermediate value between the luminance values based on the threshold values of both adjacent masks, the luminance values of the remaining luminance value-indeterminate pixels are shown in FIG. As described above, the luminance value and the distance (L1, L2, or L3, L3, L3) of the pixel (p1, p2, or p3, p4) of the known luminance at both ends of the extension line in the fixed direction of the pixel px having the undefined luminance value.
L4) to determine the luminance value by the primary interpolation method, or to calculate the luminance value of the pixel whose luminance value is undetermined by using both extended line ends of straight lines in different directions of the pixel whose luminance value is not determined in different directions. Determined by the average value of the luminance values obtained by the primary interpolation method using the luminance value and the distance of the pixel of the known luminance in the above, or the luminance value of the remaining luminance value undetermined pixel is determined by the luminance value undetermined pixel. Are determined by surface interpolation values performed using three known pixels. Thereby, a reproduced image as illustrated in FIG. 18 in which the luminance values of all the pixels are determined is obtained.

[0056]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram of a multi-dimensional multi-valued image compression / expansion system according to the present invention. FIG. 1 is a block diagram on the compression side in the multidimensional multilevel image compression / expansion system of the present invention, and FIGS. 2 to 4 show signals provided on the compression side in the multidimensional multilevel image compression / expansion system shown in FIG. 5 is a block diagram of a decompression side in the multidimensional multilevel image compression / expansion method of the present invention, and FIGS. 6 to 24 and FIGS. 26 to 29 are multidimensional multilevel image compression / expansion. FIG. 25 is a block diagram showing a configuration example of a contour tracer. In FIG. 1 showing a configuration example on the compression side in the multidimensional multilevel image compression / expansion method of the present invention, reference numeral 1 denotes a signal source of an image signal to be subjected to multidimensional multilevel image compression / expansion.

As the image signal source 1, for example, an image pickup device (TV camera) for generating an image signal, a VTR,
Others can be used. 2 is an analog-to-digital converter, 3 is a color memory, 4-1, 4-2,... 4-n
Is a signal processing unit. The specific configuration of each of the signal processing units 4-1, 4-2,... 4-n is, as exemplified in the signal processing unit 4-1, a predetermined luminance level information extraction unit 5 ( FIG. 2 shows a specific configuration example of the predetermined luminance level information extraction unit 5, a binary memory 6, and a contour tracer 7 (a specific configuration example of the contour tracer 7 is shown in FIG. 3). ), A contour line address list 8, and a polygon approximate address list 9 (a specific configuration example of the polygon approximate address list 9 is shown in FIG. 4). Reference numeral 10 denotes a multiplexer, 11 denotes an encoding unit and a transmission unit, and 12 denotes a transmission line (or recording medium).

FIG. 5 shows an example of the configuration on the expansion side in the multidimensional multi-valued image compression / expansion method of the present invention.
Reference numeral 12 denotes a transmission line (or recording medium) 12 shown on the compression side in the multidimensional multilevel image compression / decompression system of the present invention shown in FIG. The image information in a highly efficient compressed state via the transmission line (or recording medium) 12 is given to a reception decoder 13. 14-n are signal processing circuits having the same configuration, and a specific configuration example is the signal processing circuit 14.
-1. In the signal processing circuit 14-1, 15 and 16 are end point address registers, 17 is a local image memory, 18 is a polygon interpolation mask generator, 19 is a multi-value decision operator, 20 is a skeleton luminance determiner, and 21 is a luminance interpolation plane. It is an arithmetic unit. Also, in FIG.
3 is an image memory, 25 is a video signal generator, and 26 is a monitor receiver.

In the configuration on the compression side in the multi-dimensional multi-valued image compression / expansion system of the present invention shown in FIG. And supplies it to the analog-to-digital converter 2. The image signal source 1 has any configuration as long as it is a video signal generating device that can generate image information to be compressed and expanded by the multidimensional multilevel image compression and expansion method of the present invention. Can also be used. The image signal source 1 shown in FIG.
It is said that it has a configuration capable of generating a primary color signal.
The three primary color signals generated from the image signal source 1 are supplied to the analog-to-digital converter 2.

In the analog-to-digital conversion unit 2, a predetermined number of pixels (for example, the horizontal and vertical directions of the image) are obtained for each of the horizontal and vertical directions of the image for the luminance signal for each image obtained from the three primary color signals supplied thereto. A digital luminance signal Y having a predetermined number of bits (for example, 8 bits) is generated and output for each pixel in a decomposed state of 512 pixels in the direction and 480 pixels in the vertical direction of the image. Are supplied to the signal processing units 4-1, 4-2,... 4-n. As for the chroma signal, a digital color difference signal corresponding to the pixel is generated and supplied to the color memory 3.

... 4-n to which the digital luminance signal Y is supplied are shown in the dashed-dotted frame of the signal processing unit 4-1. However, in the extraction unit 5 of the predetermined luminance level information provided in each of the signal processing units 4-1, 4-2,..., 4-n, each of the signal processing units 4-1, 4-2,. For each 4-n, the image information to be compressed is binarized and output at a predetermined different luminance threshold, and the pixel address is also output. Each of the above-mentioned signal processing units 4-1, 4-2... 4
FIG. 2 shows an example of the configuration of the predetermined luminance level information extraction unit 5 provided for the digital luminance signal Y from the analog / digital conversion unit 2 to the input terminal 5a. The input terminal 5b is supplied with a clock signal from a control device (not shown).

That is, the digital luminance signal Y supplied to the predetermined luminance level information extraction unit 5 is compared with a specific luminance threshold supplied from the luminance threshold setting unit 52 in a comparator (magnitude comparator) 51. hand,
A binary output of the digital luminance signal Y is sent from the comparator 51 to the output terminal 5c. The above-described luminance threshold setting unit 52
, A predetermined luminance threshold is set in a binary number by a ROM, a DIP switch, a fuse array or the like. The brightness threshold set in the brightness threshold setting unit 52 in the predetermined brightness level information extraction unit 5 provided in each of the signal processing units 4-1, 4-2,..., 4-n is, for example, FIG. The luminance value is a predetermined luminance value when the multi-dimensional multi-valued image compression / expansion method is performed, such as the first to eighth threshold values illustrated in FIG. The comparator 51 sends a binarized output of the comparison result at the moment when the clock signal is given to the output terminal 5c.

Each of the aforementioned signal processing units 4-1, 4-2,...
The binarized output of the digital luminance signal output from the extraction unit 5 of each predetermined luminance level information in 4-n is:
The signals are stored (stored) in a binary memory 6 provided in each of the signal processing units 4-1, 4-2,..., 4-n. The binary output of the digital luminance signal stored in the binary memory 6 is an address specified by a binary memory address output from an address counter 72 (see FIG. 3) provided in the contour tracer 7. And is supplied to the determiner 71 of the contour tracer 7 (see FIG. 3).

The binarized output of the digital luminance signal output from the predetermined luminance level information extraction unit 5 in each of the signal processing units 4-1, 4-2,... 20 is a binarized signal in a state binarized at a specific luminance threshold, for example, the state illustrated in FIG. 20, and the binarized signal read from the binary memory 6 is:
In the decision unit 71 (see FIG. 3) provided in the contour tracer 7, first, the binarized pixel group as shown in FIG. 20 is read from the upper left corner of the screen in the raster scanning order. When scanning is started and scanning is performed, and the first boundary is found, tracking of the contour of a pixel having a specific luminance value is started with the point as a starting point, so that a high-luminance surface is on the left side in the tracking direction. When the tracking start direction coincides with the horizontal scanning direction (CW: clockwise),
When the tracking start direction matches the vertical scanning direction (CCW ...
(Counterclockwise) to trace the contour of the pixel and send an output signal (address strobe) of contour information to the output terminal 7c (see FIG. 3).

The information H and V of the step history (change in the position in the horizontal scanning direction, change in the position in the vertical scanning direction) in tracking the contour of the pixel having the specific luminance value is transferred to the address counter via the multiplexer 73. 72. Thus, the address counter 72 sequentially outputs the tracking address sequence as a binary memory address output to the output terminal 7b (see FIG. 3) based on the above-described step history,
It is sent to an output terminal 7d (see FIG. 3) as a contour address output. The contour line address to which the contour signal output signal (address strobe) output from each contour line tracer 7 in each of the signal processing units 4-1, 4-2,... 4-n and the contour line address output are supplied. In Listing 8, after storing the contour address output,
For example, as illustrated in FIG. 22, the coordinates of the specific isoluminance line passing point are moved to the center position of each pixel constituting the specific luminance boundary.

Each of the signal processing units 4-1, 4-2,...
The address output of the center position of each pixel constituting the specific luminance boundary output from each contour line address list 8 in 4-n is output to each of the signal processing units 4-1, 4-2,. It is supplied to the square approximate address list 9. In FIG. 4 showing a specific configuration example of the polygon approximate address list 9, an input terminal 9 a has address information of the center position of each pixel constituting a specific luminance boundary output from the contour address list 8. Is entered. Then, the address information of the center position of each pixel constituting the specific luminance boundary supplied to the input terminal 9a of the polygon approximate address list 9 is supplied to the end point address register 91 and the shift register 93. The output of the end point address register 91 is supplied to the interpolation address calculator 92.
The address information of the first storage section in the shift register 93 is also given.

In the interpolation address calculator 92, the horizontal scanning direction of the image is now set to the Xh axis, and the vertical scanning direction is set to the vertical scanning direction in order to compare the distance between each of the above-described feature points and the interpolation straight line with a simplified value for comparison. When the Yv axis is set, each of the feature points and the Xh axis in the horizontal scanning direction (or the Yv axis in the vertical scanning direction)
The address for comparing the address value of the Yv axis in the vertical scanning direction (or the address value of the Xh axis in the horizontal scanning direction) on the interpolation straight line at the above address is calculated. The switching of the corresponding axis is performed at the intersection angle between the interpolation straight line and the horizontal scanning direction Xh axis, and the threshold value is set to 45 degrees on the assumption that one pixel is a square. That is, when the absolute value of the intersection angle between the interpolation straight line and the horizontal scanning direction Xh axis is 45 degrees or less (or other than 45 degrees), the first accumulation section in the shift register 93 is sequentially shifted. The contour address information which is input and is sequentially held for the successive storage sections in the shift register 93 and the interpolation address group at the position corresponding to the horizontal scanning direction Xh axis (or vertical scanning direction Yv axis) address ( The end point address and the sequential address information in the first accumulation section in the shift register 93 are supplied to the register 94 as an interpolation address.

Then, the sequential contour pixel addresses sequentially stored in the shift register 93 and the register 9 are stored.
The interpolation addresses sequentially stored in FIG. 4 are those corresponding to each other as shown in FIG.
C3... Cn. Each comparison extractor C described above
2, C3... Cn, when the absolute value of the difference obtained as a result of comparing the contour pixel address supplied thereto and the register interpolation address exceeds a predetermined value, the output is set to the feature point address. The address value of the feature point is given to the register 95 and stored in the feature point address register 95. At the same time, the address value of the feature point is given to the end point address register 91 as a new end point address value.

The address information of the center position of each pixel constituting the specific luminance boundary supplied from the contour line address list 8 is input to the input terminal 9a in the polygon approximate address list 9 illustrated in FIG. Assuming that the operation in the polygon approximate address list 9 is started, the above-described tracking start point address information is stored in the end point address register 91 and at the same time is stored in the first storage section 1 of the shift register 93. It is memorized. The above-mentioned polygon approximate address list 9 is supplied with the sequential contour address group output from the above-mentioned contour address list 8, so that each of the storage sections 1, 2, 3,. The stored contents shift one after another.

However, the above-mentioned end point address register 9
1 has not yet changed, so that the interpolation address calculator 92 stores the first storage section 1 in the shift register 93 supplied from the first storage section 1 in the shift register 93 described above. A linear interpolation value address group in which only the contents are updated is output. As described above, each time a group of linear interpolation value addresses is output, each of the storage sections 1, 2,... N in the shift register 93 and the horizontal (or vertical) An interpolated value address group corresponding to the direction address is output to the accumulation sections 2, 3, 4,... N of the register 94.

Each of the storage sections 2, 3, 4,... N in the shift register 93 and the storage sections 2,
, C3,... Cn, respectively, from the comparison extractors C2, C2, C3.
2, C3... Cn, two pieces of input information individually supplied from the respective storage sections 2, 3, 4... N of the shift register 93 and the storage sections 2, 3, 4,. The absolute value of the difference (corresponding to the distance between the contour line and the interpolation straight line on the screen) exceeds the predetermined threshold value, and the point of the contour line address is used as a feature point. Store.

The above-mentioned multiple comparison and extraction units C2, C2, C3
... When the feature point information is simultaneously output from a plurality of comparison / extractors in Cn, the contour line address closer to the address value stored in the endpoint address register 91 is adopted as a new endpoint address. Then, the address value is stored in the feature point address register 95, and the new endpoint address is stored in the endpoint address register 91. In the above case, even if a plurality of feature point addresses disappear, it is not necessary to restore them as feature point addresses.

Each of the signal processing units 4-1 and 4-1 shown in FIG.
The feature point address group for each of the different luminance thresholds individually output from 4-2... 4-n is supplied to the encoding unit and the transmission unit 11 via the multiplexer 10. The chroma signal component is also supplied from the color memory 3 via the multiplexer 10 to the encoding unit 11. High-efficiency coding is performed and transmitted (or recorded) to the receiving side (or reproducing side) via the transmission line (or recording medium) 12.

In the above-described embodiments, the detection of the feature point address group for each of the different brightness threshold values is performed based on the contour line from the detected feature point pixel and the contour plane in the case of three-dimensional brightness information. According to｝, with respect to a virtual straight line between a pixel traced in a certain direction and a pixel between the two pixels, a pixel indicating a distance exceeding a certain threshold distance is determined to be a feature point. However, in the implementation of the present invention, the detection of the feature point address group for each different luminance threshold is performed by using the luminance function of the contour line (contour line) divided by the three-dimensional luminance information. In this case, the pixel at the local maximum point of the curvature of the isoluminance plane｝, or the contour line described above, or the contour plane に は in the case of three-dimensional luminance information, exceeds a predetermined threshold angle. The pixel in that case is determined to be a feature point, and the feature point is determined. May be in, it may be contour tracer 7 configuration aspects already described for FIG. 25 is used as a contour tracer 7.

Next, a description will be given of a case where the luminance function of the original image is expanded and restored with reference to a block diagram on the expansion side in the multidimensional multi-valued image compression and expansion system of the present invention. FIG.
, 14-n decodes the highly efficient coded signal supplied thereto by the transmission path (or recording medium) 12 to the signal processing circuits 14-1, 14-2,..., 14-n. give. A number of signal processing circuits 14-1, 14-2,..., 14-n provided on the extension side
A large number of signal processing units 4 provided on the compression side in the multidimensional multilevel image compression / decompression system of the present invention described above with reference to FIG.
-1, 4-2,..., 4-n, and the signal processing circuits 14-1, 14-2,.
.. 14-n are used for image processing of an image signal binarized using different specific luminance thresholds. A specific configuration example of the above-mentioned many signal processing circuits 14-1, 14-2,..., 14-n is shown in the signal processing circuit 14-1.

In the above-mentioned reception decoder 13, a large number of signal processing units 4-1, 4-2,..., 4 provided on the compression side in the multi-dimensional multi-level image compression / expansion system of the present invention described above.
-N, among the high-efficiency coded signals compressed for the image signal binarized using the different specific luminance thresholds, the high-level of the image signal binarized using the specific luminance thresholds is used. The efficiency coded signals are distributed and supplied to the corresponding signal processing circuits 14-1, 14-2,..., 14-n. Each of the aforementioned signal processing circuits 14-1, 1
In 4-2,..., 14-n, the image signal binarized using the different specific luminance thresholds is binarized using the specific luminance threshold of the compressed high efficiency coded signal. The signal processing for the highly efficient coded signal of the image signal is performed. The signal processing operation performed in each of the signal processing circuits 14-1, 14-2,. In the description, the signal processing circuit 14-1 will be specifically described.

When the feature point addresses sequentially supplied from the reception decoder 13 to the signal processing circuit 14-1 are stored in the end point address register 15 and the end point address register 16, the end point address register 15 and the end point The feature point address is supplied from the address register 16 to the polygon interpolation mask generator 18 as both end point addresses. The polygon interpolation mask generator 18 calculates the address of the linearly interpolated pixel at the two end points given thereto, and writes the address value in the multi-port local image memory 17 with the designated luminance.

Next, when a new feature point address is supplied from the reception decoder 13 to the signal processing circuit 14-1, the new feature point address is stored in the endpoint address register 15 as a new endpoint address. The endpoint address previously stored in the endpoint address register 15 is moved to the endpoint address register 16 and stored. Then, the polygon interpolation mask generator 18 calculates the address of the linear interpolation pixel at the two end points given thereto, and writes the address value in the local image memory 17 of the multiport with the designated luminance. The polygon interpolation mask generator 18 performs the above-described operation each time a new feature point address is supplied from the reception decoder 13 to the signal processing circuit 14-1. Is calculated, and the sequential addresses are written in the local image memory 17 at the designated luminance.

When the closed curve is completed by the above-mentioned interpolation line, the polygon interpolation mask generator 18 applies the inside of the closed curve in the local image memory 17 with the designated brightness to obtain a mask of a specific brightness level. Generate. In this manner, for each threshold value for obtaining an isoluminance line for each specific luminance value, a bright region having the above-described isoluminance line as a boundary is filled with a bright symbol, and a dark region having the above-mentioned isoluminance line as a boundary. A region can be filled with a dark sign to create a mask with a light and dark binary sign. Each of the signal processing circuits 14-1, 14-2,..., 14-n specifies a specific one of the high-efficiency coded signals compressed with respect to the binarized image signal using different specific luminance thresholds. Since the signal processing is performed on the high-efficiency coded signal of the binarized image signal using the appropriate luminance threshold, each of the signal processing circuits 1
By the signal processing operation performed in 4-1, 14-2,..., 14-n, the bright areas having the boundary of the equal luminance line for each different specific luminance value as a boundary are filled with a bright symbol, and A plurality of masks can be generated as illustrated in FIG. 6 in which a dark area as a boundary is filled with a dark symbol.

After the above operation in the polygon interpolation mask generator 18 is completed, the multi-level decision operator 19 and the skeleton luminance determiner 20 are connected to the local image memory 1 described above.
7 is operated to change the luminance plane to multi-value luminance. That is, as described above, in order to improve the quality of the reproduced image, the luminance value of the interpolation end point is not used as the intermediate value of the simple category as described above in the interpolation reproduction of the luminance value. , The masks are arranged in the order of the magnitude of the luminance value according to the threshold value, and the area between adjacent masks is
After filling with an intermediate value of the luminance values based on the threshold values of the two adjacent masks described above, the relationship between the luminance value of each pixel and the luminance values of eight surrounding pixels each having the pixel as a central pixel is as follows: When the brightness value of the surrounding pixels is equal to the brightness value of the center pixel, the brightness value of the center pixel is undecided. (2) When the luminance value of the surrounding pixels is not higher than the luminance value of the central pixel, the luminance value of the central pixel is set to (the intermediate value of luminance values by the threshold values of both adjacent masks-a) (where a is a positive value). Value). (3) When the luminance value of the surrounding pixels is not lower than the luminance value of the central pixel, the luminance value of the central pixel is set to (the intermediate value of luminance values by threshold values of both adjacent masks + a). (4) When the relationship between the brightness value of the surrounding pixels and the brightness value of the center pixel is other than the above-described relationships (1) to (3), the intermediate value between the brightness values of the threshold values of both adjacent masks and I do. This is determined according to which of the conditions (1) to (4) is satisfied, so that the quality of the reproduced image can be improved.

In the case of the undefined brightness value area, a center line (skeleton) of the area, that is, a group of points equidistant from the end of the area is extracted, and the brightness value of the pixel on the center line of the area is determined by the two masks adjacent to each other. The skeleton luminance determiner 20 is set so as to set an intermediate value between the luminance values based on the threshold (see FIGS. 13 to 16).
Operates to write the determined brightness value in the local image memory 17 described above. When the above-described brightness determination operation by the multi-value determination operator 19 and the skeleton brightness determiner 20 is completed, the brightness interpolation plane calculator 21 calculates the brightness value of the undetermined brightness pixel, for example, as shown in FIG. Intermediate level ",
A luminance value having a luminance value relationship such as “intermediate luminance−” and “intermediate luminance +” is determined, or a luminance value of a pixel on the region center line is determined by a threshold value of two adjacent masks. As shown in FIG. 19, the pixels of known luminance at both ends of the extension line in the fixed direction of the above-mentioned pixel px of undetermined luminance value, as shown in FIG. The luminance value is determined by the primary interpolation method using the luminance value of (p1, p2, or p3, p4) and the distance (L1, L2, or L3, L4), or the luminance of the above-mentioned pixel whose luminance value is undetermined The average value of the luminance values obtained by the primary interpolation method using the luminance values and the distances of the pixels of the known luminance at both ends of the extension line of the straight line in each fixed direction in the different directions of the pixels whose luminance values are not determined Or the remaining luminance value as described above The luminance value of the undetermined pixel is determined by a surface interpolation value performed using three pixels whose luminance values near the pixel whose luminance value is undetermined are known.

Each of the signal processing circuits 14-1, 14-2,..., 1
At 4-n, the image signals expanded by performing the signal processing are stored in the image memories 22 and 23. The two image memories 22 and 23 operate in such a manner that the writing operation and the reading operation are alternately and sequentially repeated. The image signals read from the image memories 22 and 23 are converted by the video signal generator 24 into video signals according to a specific scanning standard television system.
5 and the reproduced image is displayed on the display surface of the monitor receiver.

[0083]

As apparent from the above description, the multi-dimensional multi-level image compression / expansion method of the present invention provides:
Move the specific isoluminance line passing point coordinates to the center position of each pixel constituting the specific luminance boundary obtained by tracking the contour of the pixel for each predetermined specific luminance value in the luminance function of the image,
A feature point that satisfies a specific condition for the specific isoluminance line is obtained, and the position and the brightness value of the feature point of the image are transmitted, recorded, restored, and the like as information obtained by efficiently compressing the original image information. Conventionally, when a reproduced image is obtained based on a specific luminance boundary obtained by tracing the contour of a pixel, a black pixel of one pixel is generated on a specific side with respect to the tracing direction. The point that caused the deterioration of the point is that the specific isoluminance line passing point coordinates are moved to the center position of each pixel constituting the specific luminance boundary obtained by tracking the contour of the pixel for each specific luminance value in the present invention. By doing so, it is possible to improve satisfactorily, and the information that has been efficiently compressed into the information of the position and the luminance value of the feature point of the image that meets the specific condition obtained for each specific isoluminance line,
When obtaining a reconstructed image by stretching, for each threshold value to obtain an isoluminance line for each specific luminance value, a bright area bordered by the isoluminance line is painted with a bright symbol, and the isoluminance line is The dark area bordered by is filled with a dark symbol to make it light and dark 2
Creates a mask using value symbols, approximates the isoluminance line (or isoluminance surface) to a polygon (or polygon), and approximates the polygon (or polygonal approximation) with a straight line connecting the feature points during expansion. Even when the luminance information of the pixels other than the characteristic points is determined by the interpolation plane (or the interpolation solid) determined by a plurality of nearby characteristic points on the line (or the equiluminance plane), ) Does not cause the luminance order to be inverted, and therefore, does not cause an abnormal state in the luminance distribution in the image to cause deterioration of the image quality. After the area between adjacent masks is filled with an intermediate value of the luminance values based on the threshold values of the two adjacent masks, the luminance value of each pixel is Between the luminance values of the surrounding eight pixels around the pixel, respectively, (1) the luminance values of the surrounding pixels, when equal to the luminance value of the center pixel, and undecided luminance value of the center pixel. (2) When the luminance value of the surrounding pixels is not higher than the luminance value of the central pixel, the luminance value of the central pixel is set to (the intermediate value of luminance values by the threshold values of both adjacent masks-a) (where a is a positive value). Value). (3) When the luminance value of the surrounding pixels is not lower than the luminance value of the central pixel, the luminance value of the central pixel is set to (the intermediate value of luminance values by threshold values of both adjacent masks + a). (4) When the relationship between the brightness value of the surrounding pixels and the brightness value of the center pixel is other than the above-described relationships (1) to (3), the intermediate value between the brightness values of the threshold values of both adjacent masks and I do. It is determined according to which of the conditions shown in the above (1) to (4) is satisfied, and a region center line (skeleton) is extracted for an undetermined brightness value region (unknown brightness value region). ,
After setting the luminance value of the pixel on the area center line as an intermediate value between the luminance values based on the threshold values of the two adjacent masks, the luminance value of the luminance value undetermined pixel (luminance value undetermined pixel) which remains is determined by the luminance value described above. The luminance value is determined by the primary interpolation method using the luminance value and the distance of the pixel of the known luminance at both ends of the extension line in the fixed direction of the undetermined pixel, or the luminance value of the luminance value undetermined pixel, The luminance value is determined by the average value of the luminance values obtained by the primary interpolation method using the luminance values and the distances of the pixels of the known luminance at both ends of the extension line of the straight line in the fixed direction in the different directions of the pixels for which the luminance value is not determined. Or the luminance value of the remaining luminance value undetermined pixel is determined by a surface interpolation value obtained by using three pixels whose luminance values in the vicinity of the pixel whose luminance value is undetermined are known. Compression of From the image data is than a multi-tone reproduction image is obtained, according to the present invention are conventional problems described above can be satisfactorily solved.

[Brief description of the drawings]

FIG. 1 is a block diagram on the compression side in a multidimensional multilevel image compression / expansion method of the present invention.

FIG. 2 is a block diagram illustrating a configuration example of a predetermined luminance level information extraction unit in a compression-side signal processing unit.

FIG. 3 is a block diagram illustrating a configuration example of a contour tracer in a compression-side signal processing unit.

FIG. 4 is a block diagram of a polygon approximate address list in a signal processing unit on the compression side.

FIG. 5 is a block diagram on the decompression side in the multi-dimensional multi-level image compression / decompression method of the present invention.

FIG. 6 is a plan view for explaining a mask of luminance values having different threshold values.

FIG. 7 is a distribution diagram of luminance values of each pixel of an image.

FIG. 8 is a plan view used to explain four states defined by the relationship between the luminance value of one pixel in an image and the luminance values of eight pixels surrounding the one pixel.

FIG. 9 is a distribution diagram of luminance values of respective pixels of an image.

FIG. 10 is a three-dimensional diagram used to explain a specific state determined by the relationship between the luminance value of one pixel in an image and the luminance values of eight pixels surrounding the one pixel.

FIG. 11 is a diagram used to explain image expansion.

FIG. 12 is a diagram for explaining masks of luminance values having different threshold values and intermediate levels, high levels, and low levels.

FIG. 13 is a diagram illustrating a distribution example of luminance values of pixels under specific conditions in an image.

FIG. 14 is a diagram illustrating a distribution example of a skeleton in an image.

FIG. 15 is a diagram illustrating a distribution example of luminance values of pixels under specific conditions in an image.

FIG. 16 is a diagram illustrating a distribution example of luminance values of respective pixels of a reproduced image in a decompression process.

FIG. 17 is a diagram illustrating a distribution example of luminance values of pixels of a reproduced image in a decompression process.

FIG. 18 is a diagram illustrating a distribution example of luminance values of each pixel of an expanded reproduced image.

FIG. 19 is a diagram used to explain an example of determining a luminance value of an unknown pixel.

FIG. 20 is a plan view illustrating a binary image to be compressed;

FIG. 21 is a plan view showing the outline of a binary image to be compressed.

FIG. 22 is a plan view showing a state in which the outline of a binary image to be compressed has been moved so as to pass through the pixel center.

FIG. 23 is a plan view for explaining polygon approximation of a binary image to be compressed.

FIG. 24 is a plan view illustrating a pixel distribution of an expanded reproduced binary image.

FIG. 25 is a block diagram illustrating a configuration example of a contour tracer.

FIG. 26 is a diagram used to explain a method of detecting a luminance contour line and a method of determining a feature point.

FIG. 27 is a diagram for describing reproduction of a luminance function.

FIG. 28 is a diagram for describing reproduction of a luminance function.

FIG. 29 is a diagram illustrating a surface interpolation operation for reproducing a luminance function.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Signal source of the image signal to be subjected to multi-dimensional multi-valued image compression / decompression, 2 ... Analog-to-digital conversion unit, 3 ... Color memory, 4-1 to 4-n ... Signal processing unit, 5 ... Predetermined luminance level Information extraction unit, 6 binary memory, 7 contour line tracer, 8 contour line address list, 9 approximate polygon address list, 10 multiplexer, 11 encoding unit and transmission unit, 12 transmission line (or recording) 13) reception decoder, 14-1, 14-2 to 14-n ... signal processing circuit, 15, 16, 91 ... end point address register, 17
... local image memory, 18 ... polygon interpolation mask generator, 1
9: Multi-value decision operator, 20: Skeleton luminance determiner, 21: Luminance interpolation plane calculator, 22, 23, 27: Image memory, 24: Video signal generator, 25: Monitor receiver,
28-30: Address generator, 31-33: Brightness determiner, 34-36: Brightness value generator, 37-39, 93, 9
4 ... shift register, 40 to 42 ... feature point determiner, 43
... alignment circuit, 71 ... determiner, 92 ... interpolation address calculator, 95 ... feature point address register,

Continuation of the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) G06T 9/00-9/40 H04N 1/41-1/419 H04N 7/24-7/68

Claims (7)

(57) [Claims] 1. A feature point that satisfies a specific condition with respect to an equal luminance line set for each predetermined specific luminance value in a luminance function of an image is obtained, and the position of the characteristic point of the image and the luminance value are determined. In the multi-dimensional multi-level image compression / expansion method used for transmission, recording, and image restoration, the center position of each pixel constituting a specific luminance boundary obtained by tracing the contour of a pixel having a specific luminance value at the time of compression. A multidimensional multilevel image compression / expansion method in which the coordinates of a specific isoluminance line passing point are moved. 2. Finding a characteristic point that meets a specific condition for an isoluminance line set for each predetermined specific luminance value in a luminance function of an image, and determining a position of the characteristic point of the image and a luminance value. In the multi-dimensional multi-level image compression / expansion method used for transmission, recording, and image restoration, at the time of decompression, the above-mentioned equi-luminance line is used as a boundary for each threshold for obtaining an iso-luminance line for each specific luminance value. A multi-dimensional multi-level image compression / expansion method in which a bright area is painted with a bright symbol, and a dark area bordered on the isoluminance line is painted with a dark symbol to create a mask using a bright / dark binary symbol. 3. A feature point that meets a specific condition with respect to an isoluminance line set for each predetermined specific luminance value in a luminance function of an image is obtained, and the position of the characteristic point and the luminance value of the image are determined. In the multi-dimensional multi-level image compression / expansion method used for transmission, recording, and image restoration, at the time of decompression, the above-mentioned equi-luminance line is used as a boundary for each threshold for obtaining an iso-luminance line for each specific luminance value. A bright area is painted with a bright symbol, and a dark area bordered by the isoluminance line is painted with a dark symbol to create a mask with a bright / dark binary symbol. After arranging and filling the area between adjacent masks with an intermediate value of the luminance values based on the threshold values of the two adjacent masks, the luminance value of each pixel is changed to the surrounding 8 with each pixel as a central pixel. Pixel luminance value (1) When the luminance value of the peripheral pixel is equal to the luminance value of the central pixel, the luminance value of the central pixel is undecided. (2) The luminance value of the peripheral pixel is the luminance of the central pixel. When the brightness value is not higher than the value, the brightness value of the center pixel is set to “the intermediate value of brightness values based on the threshold values of both adjacent masks−a” (where a is a positive value). When the luminance value is not lower than the luminance value of the central pixel, the luminance value of the central pixel is set to “the intermediate value of luminance values by the threshold values of both adjacent masks + a”. (4) The luminance value of peripheral pixels and the luminance value of the central pixel If the relationship is other than the relationships (1) to (3) above, the intermediate value of the luminance values based on the threshold values of both adjacent masks is set to the above (1) to (4). It is determined according to which of the conditions is satisfied. Extracted and multidimensional multi-valued image compression and decompression method of an intermediate value of the luminance value by the threshold of both the mask adjacent the luminance values of the pixels of the regions centerline. 4. A feature point that meets a specific condition with respect to an isoluminance line set for each predetermined specific luminance value in a luminance function of an image is determined, and the position of the characteristic point and the luminance value of the image are determined. In the multi-dimensional multi-level image compression / expansion method used for transmission, recording, and image restoration, at the time of decompression, the above-mentioned equi-luminance line is used as a boundary for each threshold for obtaining an iso-luminance line for each specific luminance value. A bright area is filled with a bright sign, and a dark area with the above-mentioned equal luminance line as a boundary is filled with a dark sign to create a mask with a light / dark binary sign. After arranging and filling the area between the adjacent masks with an intermediate value of the luminance values based on the threshold values of the two adjacent masks, the luminance value of each pixel is changed to the surrounding 8 with each pixel as a central pixel. Pixel luminance value (1) When the luminance value of the peripheral pixel is equal to the luminance value of the central pixel, the luminance value of the central pixel is undecided. (2) The luminance value of the peripheral pixel is the luminance of the central pixel. If the luminance value is not higher than the value, the luminance value of the center pixel is set to “the intermediate value of luminance values based on the threshold values of both adjacent masks−a” (where a is a positive value). When the luminance value is not lower than the luminance value of the central pixel, the luminance value of the central pixel is set to “the intermediate value of luminance values by the threshold values of both adjacent masks + a”. (4) The luminance value of peripheral pixels and the luminance value of the central pixel Is different from the above-mentioned relations (1) to (3), an intermediate value between the luminance values based on the threshold values of both adjacent masks is shown in the above (1) to (4). It is determined again according to which of the conditions is met. The core wire is extracted, and the luminance value of the pixel on the region center line is set as an intermediate value between the luminance values based on the threshold values of the two adjacent masks. The luminance value of the remaining luminance value undetermined pixel is the luminance value undetermined pixel. A multi-dimensional multi-valued image compression / expansion method for determining a luminance value by a primary interpolation method using a luminance value and a distance of a pixel having a known luminance at both ends of an extension line in a certain direction. 5. A luminance value of a remaining luminance value undetermined pixel is determined using a luminance value and a distance of a known luminance pixel at both ends of a straight line in a fixed direction in each of different directions of the luminance value undetermined pixel. 5. The multi-dimensional multi-level image compression / expansion method according to claim 4, wherein the multi-level image compression / expansion is determined by an average value of luminance values obtained by a primary interpolation method. 6. The multidimensional multivalued image according to claim 4, wherein the luminance value of the remaining luminance value undetermined pixel is determined by a surface interpolation value using three pixels whose luminance values near the luminance value undetermined pixel are known. Compression and expansion method. 7. The multi-dimensional multi-level image compression / expansion method according to claim 3, wherein a = d / 3, where d is an interval between threshold values of adjacent masks.