JP3082957B2 - Image processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method, and more specifically, to compress image data by encoding the image data and output the encoded data to a transmission medium or a recording medium. To an image processing method.

[0002]

2. Description of the Related Art FIG. 6 shows a conventional image coding apparatus, which performs analog-to-digital conversion (hereinafter, referred to as A / D) in an image A / D converter 102 input from a terminal 101 and then codes the image. The variable length compression encoding is performed in the unit 103. Then, the variable-length compressed encoded data is temporarily stored in the transmission buffer memory 104 and transmitted to the transmission path 106.
At this time, the occupation amount of the buffer memory 104 and the transmission path 1
At a transmission rate of 06, a control coefficient for controlling the data generation amount of the variable-length compressed code of the 103 encoder is generated and fed back to the encoder 103 through the filter 105. This makes it possible to transmit the compressed image data at the transmission line rate on average. The data received from the transmission path 106 is temporarily stored in the reception buffer memory 107,
It is sent to the decoding unit 108 together with the sent control coefficient,
As a result, the variable length compression encoded data is decompressed and decoded, and D
A / A converter 109 performs digital-to-analog conversion and outputs an image from terminal 110.

As shown in FIG. 6, a number of color image compression methods have been proposed for the encoding unit 103, and a so-called ADCT method has been proposed as a typical color image encoding method.

FIG. 7 shows a conceptual diagram of a configuration of an image coding apparatus using the ADCT method. The input image is data converted to 8 bits, that is, 256 gradations / color by the A / D converter 102 in FIG.
Three colors or four colors such as B, YUV, YPbPr, and YMCK. The input image is immediately subjected to two-dimensional discrete cosine transform (hereinafter referred to as DCT) in units of 8 × 8 pixels, and then linear quantization of transform coefficients is performed.
The quantization step size differs for each transform coefficient, and the quantization step size for each transform coefficient is a value obtained by multiplying an 8 × 8 quantization matrix element by K in consideration of the difference in luminosity factor for quantization noise for each transform coefficient. And Here, K is called a control coefficient. The value of K controls the image quality and the amount of generated compressed data. Table 1 shows an example of the quantization matrix element.

[0005]

[Table 1]

After quantization, a DC conversion component (hereinafter, referred to as a DC component) is one-dimensionally predicted between neighboring sub-blocks, and a prediction error is Huffman-coded.

[0007] The quantized output of the prediction error is divided into groups, the identification number of the group to which the prediction error belongs is Huffman-coded, and subsequently, which value in the group is represented by an isometric code.

[0008] AC conversion components other than DC components (hereinafter AC
Described as a component. 8) encodes this quantized output while performing zigzag scanning from a low frequency component to a high frequency component as shown in FIG. That is, a transform coefficient whose quantized output is not 0 (hereinafter, referred to as a significant coefficient) is classified into a group according to its value, and a quantization interposed between the group identification number and the immediately preceding significant transform coefficient. Huffman coding is performed by combining the number of transform coefficients (hereinafter referred to as an invalid coefficient) having an output of 0 as a group, and subsequently, which value in the group is represented by an isometric code.

[0009]

However, in such a conventional image coding apparatus, since the amount of compressed and generated information for each image is not constant, it is difficult to estimate the amount of buffer memory in FIG. Therefore, if the amount of buffer memory is increased to prevent the failure, the cost is increased. In addition, since the control coefficient is fed back, even for the same image, the control coefficient differs depending on the previous image, the image quality changes with time, and an unsightly image may appear. Furthermore, considering a recording medium such as a magnetic tape or an optical disk as a transmission path, such a function is realized because a gap between images, such as connection, search, editing, and the like, does not match a recording data management area. It was extremely difficult.

When this technique is applied to a still image system which does not have a buffer memory, has a constant control coefficient, and does not provide feedback, the time required to transmit one image cannot be specified or recorded. However, there is a disadvantage that it is not known how much capacity is required for each sheet.

SUMMARY OF THE INVENTION It is an object of the present invention to provide an image processing method capable of controlling the amount of compressed data satisfactorily.

[0012]

In order to solve the above-mentioned problems, an image processing method according to the present invention converts image data into frequency components and encodes them, and controls a control coefficient for making the data amount of the encoded data variable. An image processing method for controlling the data amount of the encoded data, wherein the first data amount generated by the first control coefficient and the second control coefficient different from the first control coefficient The second data amount is calculated, and the relationship between the first and second control coefficients and the first data amount and the second data amount is linearly approximated. An image processing method for determining a control coefficient, wherein at the time of obtaining the first-order approximation, at least one of a zero-order term and a first-order term of a first-order approximation equation is corrected.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In the embodiment of the present invention described below, N is set to a plurality of continuous integers with respect to a control coefficient for obtaining a desired amount of information by compressing one image data. A plurality of trials are performed so that the coefficient is small and the (N + 1) th control coefficient is large, and the target information amount is trially generated by the information amount trially generated by the Nth control coefficient and the N + 1th control coefficient. Characterized in that it has means for obtaining a first-order approximation based on the amount of information obtained.

Further, the apparatus is characterized in that it has means for transforming the first-order term or the zero-order term or the zero-order and first-order terms with respect to the obtained first-order approximation formula.

By making the compression generation information constant for each image by the above means, the amount of buffer memory is minimized, and it is easy to design a stable system that does not break down due to any image. Without feedback of control coefficients, the image quality is the same for the same image, and if the transmission path is a recording medium such as a magnetic tape or optical disk, functions such as connection, search, and editing can be performed. When this technology is applied to a still image system that does not have a buffer memory and does not feed back control coefficients, the time required to transmit one image is fixed, and this technique is used for recording. It is possible to provide an image processing apparatus in which the required capacity per sheet is constant.

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an image processing apparatus according to the present invention. An image input from a terminal 1 is A / D-converted at 2, and a variable-length is converted by a coding section (1) according to the so-called ADCT method. Encoded. At this time, the control coefficient K is compressed as a constant Q ₁ for one frame of the image, thereby obtaining the compressed information amount B ₁ and sending it to the arithmetic unit 5.

At the same time, variable-length coding is performed by the coding unit (2) by the so-called ADCT method. At this time, the control coefficient K is constant Q ₂ for one image frame.
To obtain the compressed information amount B ₂ and send it to the arithmetic unit 5. Reference numeral 6 delays the A / D-converted image by the image data delay unit by about one frame.

The encoding unit (0) of 7 is operated by the arithmetic unit 5 to generate Q ₁ ,
As the optimum control coefficient K = Q ₀ calculated by the linear approximation of Q ₂ , B ₁ , B ₂ , compression encoding is performed, and the compression encoded data is stored in the transmission buffer memory 8.

Reference numeral 9 denotes a transmission line, which is a transmission medium for terrestrial radio waves such as optical fibers, satellites, and microwaves or optical space for immediate transmission, and a digital VTR or D for storage transmission.
The storage medium is a tape-shaped medium such as an AT, a disk-shaped medium such as a floppy disk or an optical disk, or a solid medium such as a semiconductor memory.

The transmission rate is determined by the information amount of the original image, the compression rate, and the required transmission time, and varies from several tens of kilobits / second to several tens of megabits / second.

On the other hand, the data received from the transmission
The compressed and coded data read out from the 10 receiving buffer memories are decompressed and decoded by the optimal control coefficient Q ₀ received at 11 at the same time, and digital-analog converted at 12 by the terminal 13. Output more images.

The present invention will be described in more detail with reference to FIGS. FIG. 2 shows an example of an image to be transmitted. One image has 1280 pixels horizontally and 1088 pixels vertically.
The image is assumed to be a D-converted image. Here, the data capacity per sheet is 1,280 × 1, 088 × 8 = 11, 141, 1
20 bits, and when this is transmitted as 30 moving images per second, 11,141,120 × 30 = 334,2
A high-speed transmission path of 33,600 bits / second is required.

On the other hand, the transmission path usually has a fixed transmission rate in most cases, and an amount of information exceeding this transmission rate breaks down and cannot be transmitted. Now, assuming a transmission path of 36.0000 Mbit / s, if the redundancy other than image information such as a sync code, ID code, and parity is 5%, the transmission rate at which image information can be transmitted is 34.2000.
M bits / sec, and the amount of compressed information of one image (one frame) is 1.1400 M bits / frame. Therefore, it suffices to compress each image of one frame to 10.23% or less, and the remaining 1,140,000- (11,141,
120 × 0.1023) = 263.424 bits / frame, that is, 263.4424 × 30 = 7, 902.7
What is necessary is just to insert 2 bits / second dummy data.

Assuming that the control coefficient is a certain value, and as a result, the compressed information amount of a certain image is 10%, the image information capacity is 334, 233, 600 × 0.1 = 33, 42
At 3,360 bits / sec, 34,200,000- (3
34, 233, 600 × 0.1) = 776, 640 bits / second dummy data may be inserted.

Assuming that the control coefficient is a certain value, and as a result, the compressed information amount of a certain image is 11%, the image information capacity is 334, 233, 600 × 0.11 = 36, 76
34,200,000- (33 at 5,696 bits / sec)
4,233,600.times.0.11) =-2,565,69
It becomes 6 bits / second, which exceeds the transmission rate of the transmission path, and causes a failure.

Therefore, the target of the compression ratio is set to 10.23%,
In FIG. 1, the optimum control coefficient Q ₀ may be given to the encoding (0) of 7 so as to obtain a value close to this.

FIG. 3 is an explanatory diagram for determining the optimum control coefficient Q ₀ . Here, a case will be described in which a so-called ADCT method is used to compress and encode data to about 1/10 and transmit the data.

The encoding method is the same as that shown in FIG. 8. Now, 8 pixels in the horizontal direction × 8 pixels in the vertical direction are set as DCT sub-blocks.
Performs linear quantization of transform coefficients. The quantization step size is different for each transform coefficient, and the quantization step size for each transform coefficient is an 8 × 8 quantization matrix element shown in Table 1 in consideration of the difference in luminosity factor for quantization noise for each transform coefficient. Is multiplied by a control coefficient K. The image quality and the amount of generated data are controlled by the value of K, and are reduced to about 1/10. After quantization, the DC component is one-dimensionally predicted between adjacent subblocks as a difference value from 0 in the first DCT subblock, and the prediction error is Huffman-coded. Then, the quantum output of the prediction error is divided into groups, and first, the identification number of the group to which the prediction error belongs is Huffman-coded,
Subsequently, any value in the group is represented by an equal length code. The AC component is encoded while zigzag scanning the quantized output from a low frequency component to a high frequency component. That is, the significant coefficients are classified into groups according to their values, and Huffman coding is performed by combining the group identification number with the number of invalid coefficients sandwiched between the immediately preceding significant transform coefficients. Now, two control coefficients Q ₁ and Q ₂ are selected, and Q ₁ <Q ₀ and Q ₀
<To satisfy the condition of Q _2.

FIG. 3 shows the relationship between the control coefficient K and the amount of compressed information Y in one general image frame. This Y and K
Is represented by a function g, that is, Y = g (k).
This function g is considered to be extremely close to a log curve close to Y = g (K) = plogK + q (p and q are constants) (1)

Therefore, the control coefficient Q ₁ is encoded by the encoding unit (1) shown in FIG. ₁ to obtain a compressed information amount B ₁ .

The control coefficient Q ₂ is encoded by the encoding unit (1) shown in FIG. 1 to obtain a compressed information amount B ₂ .

In the operation unit 5 in FIG. 1, (Q ₁ , B ₁ )
A straight line connecting two points of (Q ₂ , B ₂ ), Y = aK + b (a, b
Is a constant). Result, (Equation 1)

[0033]

[Outside 1]

Transform (Equation 2)

[0035]

[Outside 2]

In FIG. 3, if B ₀ is the target compression ratio (10.23%), B ₀ is substituted for Y in Equation 2 to obtain the optimum control coefficient Q ₀ . (Equation 3)

[0037]

[Outside 3]

Actually, the amount of compressed information generated by the optimal control coefficient Q ₀ is on Y = g (K), and is B ₀ .
Equation (1) is a 1 log curve, which is a downwardly convex curve, and a straight line connecting two points on the downwardly convex curve is always located on the upper side as shown in FIG. This always indicates that B ₀ > B ₀ ′, and in any case, does not exceed the target compression ratio and does not cause a failure in the transmission path.

Of the formula 3, Q ₁ , Q ₂ , and B ₀ are constant known values in the apparatus, and it is sufficient that only B ₁ and B ₂ are constants and can be obtained by trial of encoding. Therefore, the encoding units (1) and (2) 3 and 4 in FIG. 1 need only generate the amount of compressed information.

In FIG. 1, the calculation unit 5 calculates the above equation 3, but the calculation may be performed using a CPU or the like, or a ROM or a RAM is used. A look-up table may be used.

In the above example, the relationship between the control coefficient and the amount of compressed information has been described as a log curve. However, the relationship may be different from the log curve. It may be a simple curve. This depends on the quantization method in the encoding unit, the type of encoding, and the like. However, in each case, the curve is common in that the curve is convex downward (the tangent always exists below the curve), and the above-described method of determining the control coefficient becomes effective based on such characteristics.

FIG. 4 is a diagram showing a case where the relationship between the control coefficient K and the information amount Y is special with respect to the control coefficient K.

If the information amounts when quantized by the control coefficients Q ₁ and Q ₂ are B ₁ and B ₂ respectively and the originally desired information amount is B ₀ , as shown in FIG. = AK + b, the control coefficient is Q ₀ . Then, the actual information amount B ₀ ′ when quantized by the control count Q ₀ becomes B ₀ ′> B ₀ , and a failure occurs.

[0044] Therefore the amount of information B _1, B ₂ in the resultant approximate expression Y =
By adding a predetermined information amount β (constant) to aK + b, an approximate expression Y ′ = aK + b + β is obtained, and a desired information amount B is obtained.
If the quantization is performed using the control coefficient Q _0β obtained at ₀ , the actual information amount becomes B ₀ ″. Then, B ₀ ″ <B _{0 and the} desired information amount becomes B ₀ or less. It becomes possible to quantize, compress and transmit.

In FIG. 4, since Y = g (K) is generally a downwardly convex function, the above constant β is β> 0.
It is desirable that By adding such β,
There is no possibility of failure. β may be set manually in the calculation unit 5 according to the input image, for example, or may be automatically set according to the characteristics of the image.

FIG. 5 is a diagram showing the control coefficient K and the information amount Y when the approximate expression Y = aK + b shown in FIG. 4 is transformed (corrected) into Y '= (a + α) K + b + β. As described with reference to FIG. 3, the desired information amount B ₀ and the approximate expression Y = (a + α) K
If the quantization is performed using the control coefficient Q ₀ αβ obtained by + b + β, the actual information amount becomes B ₀ ″, and if B ₀ ″ <B ₀ , the desired information amount B ₀
In the following, it is possible to perform quantization compression and transmission.

It is desirable that α also be α> 0 as in β.

According to the above-described embodiment, it is possible to make the amount of compressed generated information in each image constant by obtaining the optimum control coefficient from the first approximation of a plurality of encoding trials. Facilitates the design of a stable system that minimizes the amount of use and does not break down with any image, and provides constant image quality for the same image without feedback of control coefficients, and furthermore, When considering a recording medium such as a magnetic tape or an optical disk as a path, it is easy to realize functions such as connection, search, and editing. In addition, when this technology is applied to a still image system that does not have a buffer memory and does not feed back control coefficients, the time required to transmit one image is fixed, and the time required for recording is required for each image. It is possible to provide an image encoding device having a fixed capacity.

Further, by correcting the first-order approximation expression, quantization compression can be performed without any failure even for an image having a special relationship between the quantization control coefficient and the information amount.

The frequency transform is not limited to DCT, and other orthogonal transform such as Hadamard transform may be used.

The block size is not limited to 8 × 8 pixel blocks.

The coding method after quantization is not limited to Huffman coding, but may be another method such as arithmetic coding or run-length coding.

The approximation similar to the linear approximation is not necessarily the first-order approximation itself, but is within the scope of the concept of the present invention.

[0054]

As described above, according to the present invention, the data amount of compressed data can be controlled well.

[Brief description of the drawings]

FIG. 1 is a configuration block diagram of a first embodiment of the present invention.

FIG. 2 is a diagram showing a transmission target image according to the embodiment explaining the present invention.

FIG. 3 is a diagram showing a calculation method according to a first embodiment for explaining the present invention.

FIG. 4 is a diagram showing a relationship between a control coefficient and an information amount.

FIG. 5 is a diagram showing a relationship between a control coefficient and an information amount.

FIG. 6 is a configuration block diagram of a conventional example.

FIG. 7 is a view for explaining a general variable-length coding method.

FIG. 8 is a detailed explanatory diagram of a general variable-length coding scheme.

[Explanation of symbols]

 3 Encoding unit (1) 4 Encoding unit (2) 5 Operation unit 6 Delay unit 7 Encoding unit (0)

Continuation of the front page (58) Field surveyed (Int. Cl. ⁷ , DB name) H04N ^7/ 24-7/68 H04N 1/41-1/419 H03M 7/30 JICST file (JOIS)

Claims (2)

(57) [Claims] 1. An image processing method for converting image data into frequency components and encoding the encoded data, and controlling the data amount of the encoded data by a control coefficient for making the data amount of the encoded data variable, comprising: Calculating a first data amount generated by the first control coefficient and a second data amount generated by a second control coefficient different from the first control coefficient; An image processing method for determining the control coefficient for a desired data amount by first-order approximation of a relationship between a coefficient and the first data amount and the second data amount. Zero-order terms and 1 in the first-order approximation
An image processing method, wherein at least one of the following items is corrected. 2. The image processing method according to claim 1, wherein the correction of the first-order approximation is performed manually.