JPH07262363A - Picture element density converting device - Google Patents

Abstract

PURPOSE:To shorten the processing time of picture element density conversion by executing four-point interpolation with a valid adjacent picture element for the picture element density conversion separately for linear interpolation in (x) and (y) directions. CONSTITUTION:An adjacent picture element managing part 60 inputs and holds the picture elements of a source image adjacent to an attention picture element from a source image memory 11 through an input image buffer managing part 50. A source image managing coordinate managing part 20 holds the positions of the picture elements of that source image as an integer, and a transformed image coordinate managing part 30 holds the position of the attention picture element as a rational number. An adjacent picture element influence degree managing part 70 calculates a distance between the picture elements of the source image required for linear interpolation and the attention picture element from the data held in the source picture element coordinate managing part 20 and the transformed picture element coordinate managing part 30, and the picture element value of the attention picture element is calculated by processing at an interpolating arithmetic part 80. After this source image is completely linearly interpolated in the (x) direction, for example, the (x) direction linearly interpolated result is similarly linearly interpolated in the (y) direction, and the picture element density converted image of the source image is generated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pixel density conversion device for performing pixel density conversion on image data at an arbitrary conversion rate at high speed.

[0002]

2. Description of the Related Art Conventionally, the pixel density of image data has been converted by using a method as shown in the following (1) and (2). (1) The value of the nearest pixel is set as the value of the converted pixel. (2) The value of the converted pixel is set by the interpolation calculation with the neighboring pixel. FIG. 2 is an explanatory diagram of positions of converted pixels with respect to pixels of the original image. FIG. 2 shows an original pixel P in the original image, a converted pixel Pe in the converted image arranged in the coordinate system of the original image, and a target pixel Q of the converted pixels. In the method (1), the four neighboring pixels P of the pixel of interest Q are
(I, j), P (i + 1, j), P (i, j + 1), P
When the distances between (i + 1, j + 1) and the target pixel Q are La, Lb, Lc, and Ld, respectively, the respective distances La are
~ Find Ld. Then, for example, when the distance L1 is the minimum, the value of the pixel of interest Q is set to the value of the pixel P (i, j). As the interpolation calculation of (2), there are a logical sum method, a nine division method, a projection method, and a weight function method.

[0003]

However, the conventional pixel density conversion method has the following problems.
With the method (1), the quality of the converted image is poor. Also,
In the method (2), the image quality of the converted image is preferably a projection method using an interpolation calculation or a weighting function method. Interpolation calculation with neighboring pixels is effective as a method for performing pixel density conversion while suppressing deterioration in image quality of a converted image, but the interpolation calculation requires a large amount of calculation and requires a long processing time.

[0004]

In order to solve the above problems, the present invention is directed to a memory section for storing values of pixels forming an image, and a position with respect to a pixel in a first image read from the memory section. An original pixel coordinate management unit that holds as an integer,
A pixel density conversion unit that stores a position of a pixel of interest in the converted image obtained by converting the first image in the converted image when the converted image is mapped to the first image; Be prepared for the device. The pixel density conversion device further includes an adjacent pixel management unit that holds two pixel values of adjacent pixels in the first image read out from the memory, and a fractional decimal number stored in the converted pixel coordinate management unit. The adjacent pixel influence degree management unit that calculates the distances between two pixels in the first image adjacent to the target pixel and the target pixel by using the portion, and selects and outputs the distance, and the adjacent pixel management The value of the pixel of interest is written to the memory unit and an interpolation calculation unit that obtains the value of the pixel of interest by interpolation using the two pixel values held in the unit and the distance sent from the adjacent pixel influence degree management unit. And writing means. Then, the pixel density conversion device of the present invention uses the original image as the first image in the first processing, performs pixel density conversion in the x direction on the original image, and forms a first converted image composed of the value of the pixel of interest. In the second process after the first process, the first converted image is used as the first image, the pixel density conversion in the y direction is performed on the original image, and the second image is formed by the value of the pixel of interest.
The converted image is obtained, and the density-converted image is generated by subjecting the original image to pixel density conversion in the x and y directions.

[0005]

According to the present invention, since the pixel density conversion device is configured as described above, the pixel density conversion device performs the x-direction pixel density conversion on the original image in the first processing,
In the second processing, pixel density conversion in the y direction is performed using the first converted image as the first image, and a density converted image in which the pixel density conversion is performed in the x and y directions is generated. In the pixel density conversion in each direction, the original pixel coordinate management unit holds the position of the pixel of the first image as an integer, and the converted pixel coordinate management unit holds the position of the target pixel in the form of rational number. The adjacent pixel influence degree management unit calculates the distance between the pixel of interest required for interpolation and the pixel of the first image adjacent to the pixel of interest using the fractional part of the rational number held in the converted pixel coordinate management unit. To do. Further, the adjacent pixel management unit holds the pixel value of the adjacent pixel in the first image adjacent to the pixel of interest required for interpolation.
The interpolation calculation unit performs an interpolation calculation using the distance from the adjacent pixel influence degree management unit and the pixel value from the adjacent pixel management unit to obtain the pixel value of the target pixel. Therefore, the above problem can be solved.

[0006]

EXAMPLE FIG. 3 is a diagram showing the positions of the original pixel and the converted pixel and is an explanatory view for explaining the outline of the present example. In FIG. 3, a plurality of original pixels P in the original image and a plurality of converted pixels P * in the converted image are shown in the coordinate system of the original image.
When the original image is composed of n lines × m pixels and the converted image is composed of N lines × M pixels, the conversion rate Rx in the x direction and the conversion rate Ry in the y direction in FIG. 3 are Rx = M / m Ry = N / n The position (x, y) of each pixel P in the original image is represented by an integer value, and as shown in FIG. 3, the distance Dx in the x direction between adjacent pixels in the original image is 1, and the distance Dy in the y direction is 1. .
At this time, when the converted pixels are mapped on this original image, the position of each converted pixel generally has a decimal part. The distance dx in the x direction between adjacent pixels in the converted image is m / M, and the distance dy in the y direction is n / N.

FIG. 4 is an explanatory diagram of the linear interpolation in the x direction. The pixel density conversion using the linear interpolation in the x direction for the original image will be described with reference to FIG. In FIG. 4, two pixels P (i), P (i
+1), the pixel of interest P * (k) in the converted image, and two pixels P * (k−1) arranged side by side with the pixel P * (k) interposed therebetween.
P * (k + 1) are shown. Pixel P of original image
The coordinate Porg (i) indicating the position of (i) is set as an integer, and the coordinate Pen (k) indicating the position of the target pixel P * (k) is set as a rational number, and each pixel P for the target pixel P * (k) is set.
(I), the distance of P (i + 1) is d1, d2, and its pixel P
Pixels P * (k-1) and P * (k +) for * (k)
Let the distance of 1) be dx. If the pixel value of the pixel P (i) is P ₁ and the pixel value of the pixel P (i + 1) is P ₂ , then x
Pixel value P of target pixel P * (k) by linear interpolation in the direction
_k becomes the following formula (1). Pen (k) = {d2 / (d1 + d2)} · P ₁ + {d1 / (d1 + d2)} · P ₂ (1) Here, since d1 + d2 = 1 (2), the pixel value P _k is It becomes d2 · P ₁ + d1 · P ₂ . Further, the integer part and the decimal part of the position Pen (k) are as follows: Integer part of Pen (k) = Porg (i) ... (3) Fraction part of Pen (k) = d1 ... (4). In this way, when the coordinates indicating the positions of the original pixel and the converted pixel are expressed, the integer part of the position Pen (k) of the converted pixel is the same as the coordinate Porg (i) as shown in equations (3) and (4). At some point, the fractional part is d1. Also, d2 is 1-d
It is represented by 1. That is, the distance between the pixel of interest required for linear interpolation in the x direction and the pixels of the original image in the vicinity thereof can be easily obtained from the position of the pixel of interest in the original image.

FIG. 1 is a block diagram showing the configuration of a pixel density conversion device according to an embodiment of the present invention, and FIG. 5 is a diagram showing details of each part in FIG. This device performs linear interpolation in the x direction on the original image as the first process when performing the pixel density conversion on the original image, and in the second process, the first obtained by the result of the first process. The linearly interpolated in the y direction is performed by using the converted image of 1) as the original image. As a result, this apparatus generates a density-converted image obtained by performing 4-point interpolation on the original image. The pixel density conversion device of FIG. 1 holds a memory unit 10 for storing pixel values forming an image and a position with respect to an original pixel, for example, as an integer, and updates the position in accordance with the reading order in the memory unit 10. , The original pixel coordinate management unit 20 that adds an integer value to set the distance Dx or Dy between adjacent pixels to 1, and holds the position of the pixel of interest when the converted image is mapped to the original image in the form of a rational number. In the case of updating, the conversion pixel coordinate management unit 30 adds the distance set according to the conversion rate Rx or Ry to the rational number. The apparatus also includes a comparison unit 40 that compares the data held in the original pixel coordinate management unit 20 and the converted pixel coordinate management unit 30, and an input buffer management unit 50 that controls the reading of pixel values from the memory unit 10. ,
An adjacent pixel management unit 60 that holds two pixel values of adjacent original pixels used for linear interpolation via the input buffer management unit 50.
And the fractional part of the rational number held in the converted pixel coordinate management unit 30, the distances d1 and d2 between the original pixel adjacent to the target pixel and the target pixel are calculated, and the distances d1 and d2 are selected. Adjacent pixel influence degree management unit 7 to be transmitted
0 and. Further, in the apparatus of FIG. 1, two adjacent pixel values P ₁ and P ₂ read from the memory unit 10 and the distance d sent from the adjacent pixel influence degree management unit 70.
1, by using the d2, the interpolation operation unit 80 for determining by interpolation pixel value P _k of the pixel of interest, the pixel value P _k of the pixel of interest in the memory unit 10 is writing write means output image buffer management unit 90 and 90 are provided.

Next, the details of each part in FIG. 1 will be described with reference to FIG. The memory unit 10 includes an original image memory 11 that stores pixel values of an original image and a converted image memory 12 that stores a converted image. The original pixel coordinate management unit 20 selects a register 21 that holds the coordinates of the original pixel and whether to update the contents of the register 21 based on the comparison result of the comparison unit 40 and stores the result in the register 21. And an adder 23 that adds 1 to the value of the register 21 at the time of updating, and the original pixel coordinate management unit 20 manages the coordinates of one of the two original pixels required for the linear interpolation processing. . The converted pixel coordinate management unit 30 determines the pixel of interest P *.
A register 31 that holds a value that represents the position coordinate Pen (k) of (k) in the coordinates of the original image, and a register 3 that holds the distance dx between adjacent converted pixels in a value that corresponds to the coordinates of the original image.
2 and an adder 33 that adds the values held in the registers 31 and 32 and gives an output to the register 31. The position Pen (k) of the pixel of interest for which linear interpolation processing is to be performed is the coordinate of the original image. to manage. In FIG. 5, the register 31
MS of the data of coordinates Pen (k) held in
The B side is shown as 31-1, and the LSB side is shown as 31-2, which is supplied to the adjacent pixel influence degree management unit 70. The comparison unit 40 includes a comparator 41 that compares the values held in the registers 21 and 31 and a register 42 that holds the result of the comparator 41.

The input image buffer management unit 50 includes a comparison unit 4
Based on the comparison result of 0, the pixel value data in the x direction and the y direction of the image are continuously read from the original image memory 11 or the converted image memory 12, and the read pixel value is supplied to the adjacent pixel management unit 60. Yes, the line buffer 51 and the control circuit 52 for controlling the timing of the line buffer 51
Is equipped with. The adjacent pixel management unit 60 includes a register 61 that holds the pixel value of one of the two pixels required for the linear interpolation process, and a register 62 that holds the other pixel value.
Selector 6 which selects the data held in the register 61 and the data from the line buffer 51 and gives the selected data to the register 61
3, a selector 64 for selecting the pixel value held by the registers 61 and 62 and giving it to the register 62,
A selector 65 for selecting the pixel value held by 62 and supplying it to the interpolation calculation unit 80 is provided. Each selector 63,
64 and 65 are controlled by the comparison result in the comparison unit 40,
The adjacent pixel management unit 60 is configured to manage two pixel values in the original image necessary for the linear interpolation processing. The adjacent pixel influence degree management unit 70 uses the data 31-on the LSB side of the register 31.
Register 71 that holds 2 as the distance d1, and distance d2
Register 72 for holding data and data 31-2 from "1"
, A selector 74 for calculating the distance d2, a selector 74 for selecting the data held in the register 72 and the subtraction result of the subtracter 73, and providing the register 72,
The interpolation calculation unit 8 selects the distance data held by 1, 72
0, and a selector 75 for supplying 0. That is, the adjacent pixel influence degree management unit 70 determines the 2
The distances d1 and d2 are managed and supplied to the interpolation calculation unit 80. The interpolation calculation unit 80 accumulates the outputs of the mul 81 and a 2-input multiplier 81 for multiplying the pixel value from the adjacent pixel management unit 60 and the distance data from the adjacent pixel influence degree management unit 70. A two-input accumulator (hereinafter referred to as acc) 82 for calculating, input registers 81a and 81b for mul81, input registers 82a and 82b for acc82, a register 83 for holding the accumulation result of acc82, and a cumulative value for acc82. It is provided with a selector 84 which selects the arithmetic result and the value "0" and gives it to the input register 82b. The output image buffer management unit 90 holds the value of the converted pixel from the interpolation calculation unit 80, and the converted image memory 12
For writing to the line buffer 91,
And a control circuit 92 for controlling the write timing of the line buffer 91.

FIGS. 6 and 7 are diagrams showing pipeline stages (1) and (2) of the pixel density conversion apparatus of FIG. 5, respectively. The operation of the pixel density conversion device shown in FIGS. 1 and 5 will be described with reference to FIGS. 5 and 6 and 7. First, the execution operation of linear interpolation in the x direction will be described. It is assumed that there are conversion pixels in the conversion image as shown in FIGS. 3 and 4 with respect to the original pixels in the original image, and the given conversion rate Rx is stored in the register 32 in advance.
In addition, a coordinate Porg indicating the position of the pixel P (i) in the original image
(I) shows in the register 21 the coordinates P of the target pixel P * (k).
en (k) stores the pixel value P of the pixel P (i) in the register 31.
₁ is in the register 62, the pixel value P2 of the pixel P (i + 1) is in the register 61, and the distances d1 and d2 are two registers 7
1 and 72, respectively. Here, the data of the coordinate Porg (i) held in the register 21 is a positive integer, and among the coordinates Pen (k) held in the register 31, the MSB side 31-1 is a positive integer, The LSB side 31-2 is a decimal part. FIG. 6 shows ST1 to ST6 of six pipeline stages (1) in this apparatus. FIG. 6 shows processing of each data in each of the pipeline stages ST1 to ST6, and each unit in FIG. 5 that performs the processing. At stage ST1, the coordinate Pen (k) held in the register 31 and the coordinate Porg (i) held in the register 21 are compared by the comparator 41, and the result is set in the register 42. That is, the coordinate Pen (k) of the pixel of interest P * (k) to be processed is the pixel P.
It is determined to be between (i) and the pixel P (i + 1). At this time, of the data of the coordinate Pen (k), the MS
The B side 31-1 has the same value as the data of the coordinate Porg (i), and "eq" is set in the register 42, for example.
Of the data of the coordinates Pen (k), the LSB side 31-2 becomes the data of the distance d1. FIG. 6 shows the operation when the data on the MSB side 31-1 and the data on the coordinate Porg (i) are the same. Also, at this stage ST1, the subtractor 7
3 subtracts the distance d1 on the LSB side 31-2 from “1”,
Find the distance d2. Here, although the distances d1 and d2 are decimal values, this data is n of 2 in the process of calculation so that the data on the LSB side 31-2 can be calculated as it is in consideration of the calculation processing by the integer calculator. It is treated as a multiplied value, and a positive integer is taken out when the final operation result is obtained. On the other hand, the coordinate Pen (k) is calculated by the comparator 41.
The distance dx held in the register 32 after the processing in
Is added by the adder 33 and updated to the coordinate Pen (k + 1). The coordinate Porg (i) held in the MSB side 31-1 of the register 31 is held as it is, and the pixel value P ₁ held in the register 62 and the pixel value P ₂ held in the register 61 are also held as they are. It

At the stage ST2, the data for each of the input registers 81a and 81b of the mul 81 is selected by "eq" held in the register 42. That is, the pixel value P ₂ selected by the selector 65 is set in the register 81a, and the distance d1 selected by the selector 75 is set in the register 81b. The coordinates Pen (k) and Porg (i) held in the register 31, the pixel value P ₁ held in the register 62, and the pixel value P ₂ held in the register 61 are held as they are. At stage ST3, mul81 performs multiplication, and the multiplication result, that is, d1 × P ₂ is the input register 82a of acc82.
Is supplied to. Further, "0" is set in the input register 82b. Next, the pixel value P ₁ is supplied to the input register 81a via the selector 65, and the distance d2 is supplied to the input register 81b via the selector 75. At stage ST4, the acc82 performs addition and sets the result in the register 82b again. The mul 81 performs multiplication, and the result, that is, d2 × P ₁ is the input register 82a.
Is supplied to. At stage ST5, acc82 performs addition, and d1 · P ₂ + d2 ·
P ₁ , that is, the pixel of interest P * (k) is obtained and held in the register 83. At stage ST6, the data held in the register 83 is stored in the output image buffer 90.

FIG. 7 shows MSB side 31-1 of register 31.
The stages ST1 to ST6 in the case where the data of No. 1 and the data of Porg (i) are not the same are shown. In this case, the position Pen of the target pixel P * (k) in FIG.
(K) is not between the position Porg (i) and the position Porg (i + 1), but the position Porg (i + 1) and the position Porg (i +).
This is the case between 2). At the stage ST1 of FIG. 7, the coordinates Pen (k) held in the register 31
And the coordinate Porg (i) held in the register 21 are compared by the comparator 41, and the result is set in the register 42. That is, it is determined that the coordinate Pen (k) of the target pixel P * (k) to be processed is not between the pixel P (i) and the pixel P (i + 1). At this time, the coordinates Pen (k)
Of the data of MSB side 31-1 is coordinate Porg (i)
1 is larger than the data of. After this determination processing, the adder 23 adds 1 to the data of the coordinate Porg (i) and updates it. Then, in the register 42, for example, "no
In the data of the coordinate Pen (k), the LSB side 31-2 is the pixel P (i +) in this case.
1) and the pixel d * of interest P * (k). Further, the subtractor 73 subtracts the distance d1 on the LSB side 31-2 from “1” to obtain the target pixel P * (k) and the pixel P (i +).
The distance d2 of 2) is obtained. The coordinate Pen (k) is
After the processing in the comparator 41 is completed, the adder 33
The coordinate Pen (k + is added to the distance dx held in 2
Updated to 1). The coordinate Porg (i) held in the MSB side 31-1 of the register 31 is held as it is, and the pixel value P ₁ held in the register 62 and the pixel value P ₂ held in the register 61 are also held as they are. It

At the stage ST2, the data for each input register 81a, 81b of the mul 81 is selected by "not-eq" held in the register 42. That is, the pixel value of the pixel P (i + 2) of the new original image supplied from the input image buffer is selected by the selector 65 and set in the register 81a as the pixel value P ₂ . The distance d1 selected by the selector 75 is set in the register 81b. Then, the original pixel value P ₂ in the register 61 is selected by the selector 64 and held in the register 62. Coordinate Pe held in register 31
n (k) and Porg (i) are retained as they are. The operations in the stages ST3 to ST6 of FIG. 7 are the same as those of FIG. 6, and the linear interpolation in the x direction is realized by advancing the processing in this way. Here, if the integer part of the coordinate Pen (k) is larger than 1 with respect to the data of the coordinate Porg (i) due to the conversion rate Rx, the coordinate Pen (k)
This can be dealt with by repeatedly executing the stages ST1 and ST2 until the integer part of (k) becomes the data +1 of the coordinate Porg (i). In the linear interpolation for the y direction, the stages ST1 to ST6 shown in FIG. 6 or 7 are performed using the result obtained by the linear interpolation processing for the x direction. That is, the image obtained by the linear interpolation processing in the x direction is read as the original image, and stages ST1 to ST6 are executed. Here, for image reading, if pixels are read in the y direction, linear interpolation in the y direction can be realized.

As described above, the present embodiment is realized by separately performing the 4-point interpolation with the neighboring pixels effective for the pixel density conversion into the linear interpolation in the x direction and the linear interpolation in the y direction. , And the distance d required for linear interpolation in each direction
1 and d2 are calculated using the coordinates of the pixel of interest P * (k) set by a rational number with respect to the pixel of the original image set by an integer. Therefore, the amount of calculation in the calculation process of the pixel value P _K of the target pixel is small. Furthermore, mul8
1 and acc82, and input registers 81a, 81b, 82a, 8 for holding distance data and pixel values for them.
Since the interpolation calculation unit 80 is configured by 2b, each distance data and pixel value can be newly stored in the registers 61, 62, 71, 72 after the stage 3 of FIGS. It is possible to operate. Therefore,
The processing time required for pixel density conversion is shortened. The present invention is not limited to the above embodiment, and various modifications can be made. The following are examples of such modifications. (1) The processing was performed separately for the linear interpolation in the x direction and the y direction, but by having a plurality of devices shown in FIG.
For example, it becomes possible to simultaneously perform a plurality of linear interpolations in the x direction. Further, it is possible to simultaneously perform linear interpolation in the x direction and the y direction, and it is possible to further reduce the processing time for pixel density conversion. (2) Each unit shown in FIG. 5 constitutes the above embodiment as separate hardware, but each unit can also be realized by software using a computer.

[0016]

As described above in detail, according to the present invention, four-point interpolation with neighboring pixels effective for pixel density conversion is divided into linear interpolation in the x direction and linear interpolation in the y direction. Can be done by That is, the adjacent pixel influence degree management unit sets x
The distance required for the linear interpolation in the direction and the linear interpolation in the y direction is calculated using the position of the pixel of interest set by a rational number with respect to the pixel position of the first image set by an integer, and the interpolation calculation unit Is a configuration for performing interpolation calculation using the calculation result and the pixel value obtained from the adjacent pixel management unit. Therefore, four-point interpolation with neighboring pixels effective for pixel density conversion can be performed separately for linear interpolation in the x direction and linear interpolation in the y direction. As a result, the amount of calculation in the pixel density image conversion is reduced, and the processing time can be shortened.

[Brief description of drawings]

FIG. 1 is a configuration block diagram showing a pixel density conversion device according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram of positions of conversion pixels with respect to pixels of an original image.

FIG. 3 is an explanatory diagram showing the positions of an original pixel and a conversion pixel for explaining the outline of the present embodiment.

FIG. 4 is an explanatory diagram of linear interpolation in the x direction.

5 is a diagram showing a detailed configuration of each part of FIG.

6 is a diagram showing a pipeline stage (1) of the pixel density conversion device of FIG.

FIG. 7 is a diagram showing a pipeline stage (2) of the pixel density conversion device of FIG.

[Explanation of symbols]

 10 memory unit 11 original image memory 12 converted image memory 20 original pixel coordinate management unit 30 converted pixel coordinate management unit 40 comparison unit 50 input buffer management unit 60 adjacent pixel management unit 70 adjacent pixel influence degree management unit 80 interpolation calculation unit 90 output image Buffer management section

Claims (1)

[Claims] 1. A memory unit that stores values of pixels forming an image, an original pixel coordinate management unit that holds a position with respect to a pixel in a first image read from the memory unit as an integer, and the first unit. And a converted pixel coordinate management unit that holds the position of the pixel of interest in the converted image when the converted image obtained by converting the image is converted to the first image, and is read from the memory. Adjacent to the target pixel by using an adjacent pixel management unit that holds two pixel values of adjacent pixels in the first image and a fractional part of a rational number held in the conversion pixel coordinate management unit. An adjacent pixel influence degree management unit that calculates a distance between each of the two pixels in the first image and the target pixel, and selects and outputs the distance, and two pixels held by the adjacent pixel management unit Value and the adjacency An interpolation calculation unit that obtains the value of the pixel of interest by an interpolation calculation using the distance sent from the elementary influence degree management unit, and a writing unit that writes the value of the pixel of interest in the memory unit are provided. In the processing, the original image is used as the first image, pixel density conversion in the x direction is performed on the original image to obtain a first converted image composed of the value of the pixel of interest, and the first image after the first processing is obtained. In the second process, the first converted image is used as the first image, pixel density conversion in the y direction is performed on the original image to obtain a second converted image composed of the value of the pixel of interest, and the original image is obtained. A pixel density conversion device characterized by generating a density conversion image in which pixel density conversion is performed in the x and y directions.