JPH0459827B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image signal forming method, and is particularly applicable to converting an analog image signal into a digital code signal.

[Background technology and its problems]

For example, when trying to convert an analog video signal obtained by scanning a still image into a digital image signal, it is important to simplify the configuration for processing such as transmitting and recording the digital image signal, and to shorten the processing time. Therefore, it is desirable to be able to digitally encode still images using as few bits as possible without damaging the image information.
In order to meet these conditions, conventional methods as shown in Fig.
An image signal forming method using the PCM encoding method was used.

In FIG. 1, 1 is a still image plane, which is divided, for example, into 512 parts in the vertical and horizontal directions, thus forming 512×512=262144 pixels 2. The luminance of each pixel 2 is converted into a PCM conversion signal S1 consisting of a digital code of a predetermined number of bits, and for example, 64 luminance gradations are expressed as a 6-bit digital binary number. In principle, it is thought that the greater the number of gradation steps, the more faithful the encoding can be, but in practice, it is said that the brightness gradation of each pixel 2 can be digitally encoded with about 4 to 6 bits. .

However, in order to obtain an image signal consisting of a PCM converted signal with this number of bits, the number of bits required for the entire still image becomes enormous, and therefore the configuration and processing time of the processing system for this PCM converted signal are required. It is unavoidable that it will be large and take a long time. That is, in still image 1 in Fig. 1, 6 bits (therefore 64
To convert to a digital code signal of gradation)
An amount of information of 512×512×6=1572864 bits is required.

In order to solve this problem, an image signal forming method using a predictive coding method (DPCM method) as shown in FIGS. 2 to 5 has been proposed. This method is
This is an attempt to reduce the number of bits of a digital code signal by utilizing the characteristics of the human eye's image discrimination ability.The human eye recognizes areas where adjacent pixels have a large change in brightness compared to areas where the change in brightness is small. It has the characteristic that even if the number of gradation steps of encoding is reduced, it does not feel obtrusive. For example, pixels 4A, 4B, 4C, 4D in the first row of image 3 shown in FIG.
4E, first, as shown in FIG. 3, the pixels 4A to 4E are scanned from left to right to obtain a PCM converted signal S2 which is PCM converted according to the luminance of each pixel. In this example, pixel 4A-4
B-4C-4D-4E has gradations of "white" - "black" - "intermediate color close to black" - "intermediate color close to white" - "white", and the PCM conversion signal S2 The gradation level changes to K ₄ −K ₀ −K ₁ −K ₃ −K ₄ . This PCM conversion signal S2 is then converted into a differential signal S3 (Figure 4).
and corresponds to each pixel 4A to 4D.
Obtained from the left pixel from the PCM conversion signal S2
This is the difference value obtained by subtracting the PCM conversion signal S2. This difference signal S3 is quantized into a binary code by a tapered quantization function as shown in FIG. Coarsely quantize. Incidentally, in the case of Fig. 5, when the number of gradations of the difference signal S3 is "1", it is converted to a binary number corresponding to the decimal number "1", and when the number of gradations of the difference signal S3 is "2" and "3", it is converted to a binary number corresponding to the decimal number "1". When converted to a binary number corresponding to the decimal number "2", when the number of gradations of the difference signal S3 is "4", "5", "6"
Convert to binary number corresponding to decimal number "3", and convert to binary number corresponding to decimal number "4" when the gradation number of difference signal S3 is "7", "8", "9", "10". Convert.

In this way, the difference signal S3 becomes, for example, "10".
In this case, the encoded digital image signal only needs to have a value equivalent to "4" in decimal notation, and the number of bits of the digital image signal can be further reduced (in this case, it can be expressed with a 2-bit binary code). By the way, if it is not converted by the tapered quantization function, the difference signal
When S3 is "10", the digital image signal converts this difference signal S3 into a binary number, so a 4-bit binary code is required.

In this way, by performing tapered quantization processing that is difficult to detect by the human eye according to the conventional DPCM encoding method, a digital image with a much smaller number of bits than in the case of the PCM encoding method shown in Fig. 1 is created. I can get a signal. But this
Since the DPCM coding method fundamentally changes the gradation of each pixel of a still image, the quality of the image obtained by reproducing the changed digital image signal is inevitably degraded. Therefore, it cannot be applied when such deterioration of image quality cannot be tolerated.

[Purpose of the invention]

The present invention has been made in consideration of the above points, and
The present invention attempts to propose an image signal forming method that allows the number of bits of a digital image signal to be reduced without essentially changing the gradation of each pixel of the original image.

[Summary of the invention]

In order to achieve this purpose, the present invention includes:
Divide the pixels into a predetermined number of blocks based on the PCM conversion signal obtained from the original image, find the maximum and minimum values of the luminance gradation of the pixels included in each block, and calculate the dynamometer that is the number of gradations between the two. Difference gradation number data from a reference value determined according to the range to the luminance gradation of each pixel is obtained, and a digital image signal is formed by including this difference gradation number data and reference value data.

〔Example〕

An embodiment of the present invention will be described in detail below with reference to the drawings. Original image 11 to be converted into a digital image signal
As shown in FIG. 6, 512 pixels are arranged in the vertical direction and 440 pixels are arranged in the horizontal direction, and a black and white image is composed of 512×440=225,280 pixels as a whole.

The luminance of each pixel is converted into a 6-bit luminance code signal in the same manner as described above with respect to FIG. 1, and therefore the luminance of each pixel is represented by a 64-gradation PCM conversion signal.

Next, based on this PCM conversion signal, the original image 1
The pixels constituting 1 are divided vertically and horizontally into blocks 12 of 8 pixels, thus arranging 512 ÷ 8 = 64 blocks in the vertical direction and 440 ÷ 8 = 55 blocks in the horizontal direction. The original image is divided into 64×55=3520 blocks as a whole. Then, the maximum value M and the minimum value L of the luminance gradation Y _i of each pixel included in each block are determined, and as a result, the dynamic range D of the luminance gradation is determined by the following equation.

D=M−L...(1) This means that each block 1, as shown in FIG.
The brightness gradation of the pixels included in 2 takes any value between the minimum value L and the maximum value M, and therefore the brightness gradation of each pixel is the brightness gradation (first) based on the minimum value L.
This means that it is up to the (D+1)th gradation number. Therefore, if the luminance gradation Y _i of a pixel included in the block is expressed as the difference in the number of gradations from the minimum value L, then the luminance gradation of each pixel will be equal to the maximum dynamic range D expressed in binary numbers. This means that it can be expressed as a binary number with the number of bits d.

Therefore, the brightness gradation of a pixel included in each block is determined by the signal part representing the minimum value L of the brightness gradation in the block, and the number of difference gradations from the minimum value L of each pixel within the dynamic range D for each pixel. It can be seen that it can be expressed by dividing and representing each signal part.

In order to realize this as a digital image signal, a digital code signal S11 having the format shown in FIG. 8 is used. In other words, for each block, a minimum value signal consisting of the minimum value data of the brightness gradation is generated.
W1 is obtained, and then the pixel data bit number signal part W2 is obtained, which is data representing the number of bits of the number of differential gradations (this corresponds to the number of bits required to represent the dynamic range D), and then each A pixel data signal portion W3 is obtained by sequentially arranging pixel difference gradation number data in time series.

For example, in the format for a block of luminance distribution as shown in FIG. 7, 6-bit data representing the minimum value L is used as the minimum value signal portion W1. In this case, the dynamic range D has the range of the number of gradations represented by 5 bits, so 3-bit data representing the decimal number "5" as the binary number "101" is used as the pixel data bit number signal part W2. In addition, in this case, the number of differential gradations from the minimum value L of each pixel is expressed as a number within the dynamic range D (in other words, a 5-bit binary number), so
The pixel data signal section W3 is constructed by sequentially arranging 5-bit differential gradation number data DY _i (i=1 to 16) of each pixel in time series. Here, the pixel data signal section
Since W3 is expressed as an array of binary codes "0" and "1", it cannot express the division of data for each pixel by itself, but this division is determined by the bit number signal section.
Specified as the number of bits to be separated by W2.

Therefore, in order to reproduce the original PCM conversion signal from the digital image signal S11 in the format shown in FIG. The difference gradation number data DY _i for each pixel can be obtained by extracting the difference gradation number data DY i, and the minimum value data of the minimum value signal portion W1 may be added to this difference gradation number data DY _i .

In this way, the amount of information for each block (8 x 8 pixels) is maximized, with 6 bits (64 gradations can be expressed) as the minimum value signal part W1 and 3 bits (0 Bit~
(up to 6 bits can be specified), pixel data signal section W3
8 x 8 x 5 = 320 bits, resulting in a total of 329 bits.

Incidentally, since the information amount of the PCM conversion signal S1 in the case of FIG. 1 is 8 x 8 x 6 = 384 bits, the information amount is compressed to about 86% with the configurations shown in FIGS. 6 to 8. This means that the original data can be extracted from the compressed digital image signal.
It is possible to faithfully reproduce PCM converted signals without distorting them.

When the effect of compressing the amount of information was confirmed on video signals of still images of various contents, the following results were obtained. 69.2 [%] for monoscope images, 68.2 [%] for indoor workplace scenes, 72.6 [%] for building fire drills,
47.9 for a very monotonous announcer's upper body image
[%], 56.8 [%] in the case of a facial thrust, and 51.4 [%] in the case of a tennis player with a plain background. These experiments have confirmed that, in general, the more monotonous the image is without drastic changes in brightness, the higher the compression ratio (in other words, the less information the digital image signal has). By the way, in general, still images as a whole, such as dark images, bright images, and images with intermediate brightness, are not limited to the full range of the maximum number of gradations (64 gradations in this example), but rather to the brightness of the image. Since there are many images having brightness gradations within a corresponding predetermined range, the dynamic range D can be made small in many cases, and a considerable compression effect can be obtained.

The above image signal forming method is carried out by the image signal forming apparatus 15 having the configuration shown in FIG. In other words, the still image analog video signal S15 is
PCM converted by digital converter 16
A PCM conversion signal S16 is obtained and stored in the input data memory 17 as luminance gradation data of each pixel. This storage output signal S17 is converted into a digital image signal in the format shown in FIG. 8 by the signal processing circuit 19 formed in the microcomputer 18 together with the input data memory 17 according to the procedure shown in FIG.
Converted to S11.

That is, in step SP1 of FIG. 10, the signal processing circuit 19 reads out one block's worth of PCM conversion signals S17 from the input data memory 17 in a predetermined order, and selects the maximum value of the number of brightness gradations of pixels included in the block. M and the minimum value L are calculated and stored in an output data memory provided in the signal processing circuit 19. Next, the signal processing circuit 19 moves to step SP2 to calculate the dynamic range D (=ML), and in the next step SP3, the number of gradations representing this dynamic range D is expressed as a binary number. The number of bits d is calculated as binary code data, and this is stored in the output data memory as bit designation data of the differential gradation data of each pixel. Next, the signal processing circuit 19 moves to step SP4, reads the number of brightness gradations of the i-th (i=0) pixel included in the block from the input data memory 17, and returns the number of differential gradations DY _i ( = Y _i −L)
is calculated and stored in the output data memory. After that, the signal processing circuit 19 moves to step SP6, adds "1" to the pixel number flag i, and then judges in step SP7 whether the pixel number flag i has become the end flag i=16 of the block, and returns a negative result. When obtained, return to step SP4. Thus, the processing of steps SP4 to SP7 is repeated for the next pixel (i=1). Thereafter, the signal processing circuit 19 repeats this operation until i=16, and when it determines that i=16 in step SP7, it proceeds to step SP7.
Move to SP8 and enter the next block's signal processing program.

In this way, the output data memory of the signal processing circuit 19 stores the data constituting the minimum value signal part W1 in the format shown in FIG. 8 at step SP1, and
Furthermore, the data forming the image data bit number signal part W2 is stored in step SP3, and the data forming the pixel data signal part W3 is stored in step SP3.
It is formed and stored in SP4 to SP7, and then, when an output command is given, the signal processing circuit 19 reads it out and sends it out as a digital image signal S11.

This digital image signal S11 is transmitted to the reproducing device 21 shown in FIG. 11 (depending on the transmission distance, a transmission cable and a modem connected to the transmitting and receiving ends thereof may be inserted). The reproduction device 21 stores the transmitted digital image signal S11 in an image signal memory 23 configured in, for example, the macrocomputer 22, and stores this stored data S21 in the signal processing circuit 24.
is read out and restored to the original PCM converted signal S22, and stored in the PCM converted signal memory 25. Thereafter, when an output command is given as necessary, the signal processing circuit 24 sequentially reads out the PCM conversion signals from the memory 25 and converts them into the digital-to-analog conversion circuit 26.
In the step, it is converted into an analog image signal S23 and sent to, for example, a display device.

Note that in the above, pixel data signal section W3 (eighth
In order to obtain the differential gradation number data DY _i of each pixel composing the image (Fig. However, instead of this, the maximum value M may be selected as the reference value, and the difference from this reference value to the number of brightness gradations of each pixel may be used.

Incidentally, in a normal image, the maximum value M of brightness in an image of a certain object is determined by the intensity of the light source and the reflectance of that object, and the maximum value M of brightness within a block included in the image of that object is similar. Indicates the value. Therefore, the maximum value M is selected as the reference value, and the difference gradation number data for the luminance gradation of each pixel is obtained from this maximum value M. 2
If the bits are run-length encoded (the upper two bits have the same content based on the similarity of the maximum values mentioned above), the amount of information can be compressed. According to experiments, when 4×4=16 pixels were used as a block, the compression ratio could be improved by about 1.5%, and an image signal with a correspondingly small amount of information could be obtained.

FIG. 12 shows another embodiment in which the number of luminance gradations that can be taken within the dynamic range D is always four. In this case, each block 31 has four pixels 32 arranged in the vertical and horizontal directions, and includes 4×4=16 pixels as a whole.

Then, in the same manner as described above with respect to FIG. 6, the maximum value M and minimum value L (obtained from the 6-bit PCM conversion signal) of the luminance gradation Y _i of each pixel included in each block are determined, and the above-mentioned ( 1) Calculate the dynamic range D using the formula.

However, in this case, this dynamic range D is divided into four equal parts, and as shown in FIG.
D3 is set, and it is determined to which range of the gradation determination reference ranges D0 to D3 the luminance gradation Y _i of each pixel 32 belongs. 2 for each reference range D0, D1, D2, D3
Judgment outputs "00" to "00" are assigned decimal codes "00", "01", "10", and "11" and have a 2-bit code corresponding to the reference range to which the brightness gradation Y _i of each pixel belongs.
"11" is sent as the difference gradation number data of the pixel data signal section.

Thus, a digital code signal having the format shown in FIG. 14 is used as the digital image signal S31 in this embodiment. That is, for each block, a maximum value signal part W11 consisting of data of the maximum value M of the luminance gradation and a minimum value signal part W12 consisting of data of the minimum value L of the luminance gradation are obtained, and then the difference gradation of each pixel is obtained. A pixel data signal portion W13 is obtained by sequentially arranging several data Y _i (i=1 to 16) in time series.

In the case of this embodiment, the maximum value signal section W11 and the minimum value signal section W12 represent a reference section for obtaining the difference gradation number data DY _i of each pixel, and also represent the dynamic luminance gradation within the block. It also represents a range.

In this way, the amount of information for each block (4 x 4 pixels) can be reduced to the maximum value signal portion which becomes the reference signal.
6 bits each as W11 and minimum value signal part W12, 4 x 4 x 2 = 32 as pixel Tadar signal part W13
bit, for a total of 44 bits.

On the other hand, since the amount of information in the PCM conversion signal S1 in FIG. 1 is 4×4×6=96 bits, the configuration shown in FIGS. 12 to 14 compresses the amount of information to approximately 46%. It means that you were able to.

The digital image signal S31 shown in FIG.
To reproduce the PCM converted signal, the maximum value signal part must be
The maximum value gradation number M and the minimum value gradation number L transmitted by W11 and minimum value signal W12 are reproduced to obtain a dynamic range D (=ML), and this dynamic range D is The four judgment reference ranges D0 to D3 are reproduced by dividing them into four equal parts, and the representative values, for example, the intermediate values, of each judgment reference range D0 to D3 are divided into four gradation numbers y _i0 , y _i1 , y as shown in FIG. Calculate as _i2 and y _i3 .
Then, differential gradation number data DY _i (i=1
-16) The contents "00" to "11" are read, and the corresponding gradation numbers y _i0 to y _i4 are sequentially selected and sent out as reproduced PCM signals.

Therefore, according to this second method, the difference gradation number data corresponding to the PCM conversion signal obtained from the original image is calculated using the maximum gradation number M and the minimum gradation in units of 4 x 4 pixel blocks. By obtaining the dynamic range D based on the number L, the information amount of the image signal can be compressed.

Incidentally, when obtaining the difference gradation number data from the gradation number of each pixel,
Perform processing to apply it to one of D0 to D3, and read the code signal "00" to "11" obtained by applying it during playback, and then read the representative value y _i0 of the quadrant range.
~y Due to the process of converting to _i3 , the image quality of the playback image may be affected. However, as a result of experiments, the image quality of areas where the gradation changes relatively slowly, such as faces, is slightly degraded due to this effect, and there are problems with monoscopes, such as thin shadows appearing at the edges of lines. The texture of the material is well expressed, and in this respect the PCM conversion signal (Fig. 1)
It was confirmed that it was possible to obtain a reproduced image with a quality close to that of .

The reason for this is that if 4 x 4 = 16 pixels are considered to be one block, only four pixels are lined up in series in the vertical, horizontal, and diagonal directions. It is thought that a good approximation can be made in expressing relative brightness.

〔Effect of the invention〕

As described above, according to the present invention, the
The pixel is divided into a predetermined number of blocks based on the PCM conversion signal, the maximum and minimum values of the brightness gradation of the pixels included in each block are determined, and the dynamic range is calculated based on the number of gradations between the two. By obtaining the difference gradation number data from the determined reference value to the luminance gradation of each pixel and forming a digital image signal including this difference gradation number data and reference value data, the PCM conversion signal can be A digital image signal whose information amount is further compressed compared to the amount of information can be easily obtained.

[Brief explanation of the drawing]

Fig. 1 is a schematic diagram showing an image in a conventional image signal forming method, Fig. 2 is a schematic diagram showing the luminance distribution of the pixel, and Fig. 3 is a signal waveform showing the PCM conversion signal obtained from Fig. 2. 4 is a signal waveform diagram showing the difference signal, FIG. 5 is a curve diagram showing a tapered quantization function, FIG. 6 is a schematic diagram showing an image in the image signal forming method according to the present invention, and FIG. The figure is a schematic diagram showing the luminance gradation distribution of the block, FIG. 8 is a schematic diagram showing the format of the image signal to be formed, and FIG.
10 is a flowchart showing the signal processing procedure, FIG. 11 is a block diagram showing a reproduction device, and FIG. 12 is a block diagram showing an image signal forming device implementing the method of the present invention. FIG. 13 is a schematic diagram showing an image according to the embodiment, FIG. 13 is a schematic diagram showing a determination reference range, and FIG. 14 is a schematic diagram showing a format of an image signal to be formed. 1, 3, 11...image, 2, 4A to 4E, 32
...Pixel, 12, 31...Block, W1...Minimum value signal section, W2...Pixel data bit number signal section, W3...Pixel data signal section, W11...Maximum value signal section, W12...Minimum value Signal section, W13...
Pixel data signal section.

Claims (1)

[Claims] 1. (a) Converting an analog image signal obtained from each pixel of an original image into a PCM conversion signal expressed by luminance gradation, and (b) Converting the PCM conversion signal to a predetermined number of pixels. (c) Find the maximum and minimum values of the number of brightness gradations of pixels included in each block, (d) Calculate the reference value based on the maximum value and/or minimum value. (e) Based on the above maximum value and the above minimum value, calculate the dynamic range consisting of the number of brightness gradations between the above maximum value and the above minimum value, (f) Calculate the brightness of each pixel from the above reference value. (g) obtain a digital image signal including the reference value data representing the above-mentioned reference value, the dynamic range data representing the above-mentioned dynamic range, and the above-mentioned difference gradation data; An image signal forming method characterized by forming an image signal.