JP2005045527A - Method for compressing still image and method for compressing video signal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a video signal compressing method realizing not only improvement of the compression ratio but also reduction of the operation load for compression. <P>SOLUTION: In the case of compressing the signals constituted in frame groups in which still image frames are arrayed in time series as to completely reproduce the image signal, respective pixels (i,j) on each frame are processed in the reverse direction together for the scanning lines, a section in which the same pixel values are continued is extracted as a signal flat part, the section is masked, and after applying linear predictive error conversion to the pixels which are not masked, and the value of each pixel is converted in variable bit lengths. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to the field of video production such as video production, storage of video data materials, and relaying of location materials, especially the field of high-definition video production with higher quality than NTSC, the field of CG animation video using computer graphics, science Visualized video production field in technical simulation, video recording and playback device field using digital recording media such as CD, DVD, and semiconductor memory, storage and transmission of medical still images and moving images in electronic medical records and telemedicine, remote sensing and satellite images The present invention relates to a data compression technique that is suitable in fields where data modification is hated, such as data storage and transmission in the analysis / map information processing field.

  Conventionally, various methods are used for compressing video signals. MPEG-1, MPEG-2, MPEG-4, and the like have been put to practical use as methods for compressing and encoding video signals. Such a compression encoding method makes it possible to handle a video signal as small data, and contributes to the efficiency of data recording and transmission.

  The above-described MPEG-1, MPEG-2, MPEG-4, etc. are all referred to as lossy encoding methods and can be efficiently compressed, but there is a considerable quality degradation in decoding. The original signal cannot be reproduced completely. Therefore, in the video production field such as video production, material storage, and location material relaying, these encoding methods cannot be applied, and although inefficient, a method for storing and transmitting without compression is used. In particular, recently, the production of high-definition HDTV video has increased, and the material capacity has become enormous, which has become a problem in managing work disks.

In order to solve such a problem, a method of predicting surrounding pixels with a plurality of Markov models (see Patent Document 1) has been proposed.
JP 11-164303 A

  The conventional method has a problem that enormous calculation must be performed in order to increase the compression rate. In particular, if it is performed on an image created by CG (computer graphics) or the like with little fine change on one frame, an unnecessarily large number of calculations are performed. Accordingly, an object of the present invention is to provide a still image compression method and a video signal compression method capable of reducing the computation load for compression as well as increasing the compression rate.

  In order to solve the above problems, in the present invention, as a method of compressing the amount of information so that the video signal can be reproduced with respect to a video signal configured as a frame group in which frames that are still images are arranged in time series, A signal flat part processing stage that detects a section in which each pixel data on each frame is the same in accordance with the scanning line direction, masks the detected section as a mask area, and extracts information of the detected section as signal flat part data; The target pixel and the neighboring pixel using the pixel value of the target pixel and the pixel value of the neighboring pixel located in the vicinity of the target pixel, with each pixel existing in the non-masked area not masked in each frame as the target pixel Calculate multiple prediction error candidates for multiple prediction calculation formulas with different linear coefficients multiplied by the pixel value of, and match the optimal prediction error candidates. Adopted as a pixel prediction error, a prediction error calculation step that replaces the value of the target pixel with the prediction error, and a variable length encoding step that encodes each pixel data replaced with the prediction error with a variable length are executed. It is characterized by that.

  According to the present invention, each frame is processed in the scanning line direction to extract a signal flat portion, and pixels other than the extracted portion are calculated for each of a plurality of prediction formulas using neighboring pixels. Since the optimum value is selected as the prediction error of the pixel and the pixel recorded with the prediction error is encoded with a variable length, the portion where the same pixel value peculiar to CG etc. continues is Extracted as a signal flat part, and the other part can be recorded with a small number of bits due to prediction error and variable length coding, and as a whole, the amount of calculation is suppressed and compression with a high compression rate is performed. There is an effect that becomes possible.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
(Structure of digital video signal)
First, a digital video signal to be subjected to lossless compression in the video signal compression method according to the present invention will be described. In this embodiment, a case where a color video signal composed of three primary color signals is applied as a video signal will be described as an example. FIG. 1A is a diagram schematically showing a digital video signal handled in the present invention. In FIG. 1, the left-right direction is a time-series direction, and the time progresses as it goes to the right. The video signal shown in FIG. 1A is composed of three planes of R, G, and B in which one frame is the three primary colors of light.

In order to obtain a digital video signal as shown in FIG. 1A, first, a video is photographed with a TV camera or the like and digitized. In the case of CG or the like, the data itself is input without passing through the camera. In this way, a digital video signal is obtained. Such a digital video signal has a plurality of frames (still images) per second, and each frame has luminance values of three colors of R, G, and B. In this embodiment, it is assumed that 30 frames per second and 8 bits are assigned to each color of each pixel. For example, a digital video signal recorded for 10 seconds becomes a frame group composed of 300 frames and 900 planes.
(Compression method of the present invention)
Next, an outline of a video signal compression method according to the present invention will be described. The compression method of the present invention is executed by a computer and a dedicated software program installed in the computer. FIG. 2 is a flowchart showing an outline of a video signal compression method according to the present invention. First, a digital video signal as shown in FIG. 1A is read by a device for compression (such as a computer equipped with dedicated software). Then, the compression device starts processing. First, correlation processing between color signals is performed for each plane of each frame of the digital video signal (step S1). Specifically, based on the value of each pixel in each of the R, G, and B planes, calculation is performed according to the following [Equation 1], and each of the R, G, and B planes is converted into new X, Y, and Z planes. Reconfigure to each plane. As a result, as shown in FIG. 1B, each frame is configured by X, Y, and Z planes.
[Formula 1]
X = R−G
Z = BG
Y = G + X / 3 + Z / 3
In [Formula 1], X indicates that it is obtained by the difference calculation between G and R, and Z indicates that it is obtained by the difference calculation between G and B. In addition, since X and Z are included in the element on the right side of the third equation for obtaining Y, the first equation for obtaining X and the second equation for obtaining Z before the calculation by the third equation. Must be processed. Here, in any case, G is based on G because the luminance signal is mainly G, and the spatial resolution of the RG and BG color difference signals is about ½ of the G signal. It is. That is, when the two planes R and B are used as the difference signals X and Z, the luminance difference between adjacent pixels is reduced, and the data is suitable for compression. When the input video is monochrome, both X and Z values are “0”.

  Subsequently, the signal flat portion is processed (step S2). The signal flat portion is a portion where the signal level is a uniform value. Except for computer graphics images, the minimum value “0” of the signal level or the maximum value (eg, 255) of the signal level is a normal candidate as a uniform value. In a live-action image, such a part is generally rare, but the illumination becomes dark and the A / D converter input is below the zero calibration level (the camera signal output becomes negative when dark due to camera calibration). This may occur when the image is backlit and the A / D converter is saturated. Regardless of whether the signal level is “0” and the signal level is the maximum value, the same signal level is continuously recorded in a predetermined area (a predetermined number of pixels) in the signal flat portion. For this reason, this portion is easily compressed data. That is, in step S2, a portion where pixels having a value difference of “0” from adjacent pixels continues is separately recorded as signal flat portion data. The direction of pixel difference calculation is performed in the pixel scanning line direction. Specifically, as long as the difference between the adjacent pixel and the pixel value is “0”, the comparison with the next pixel is performed, and when there is a pixel whose difference is “1” or more, the immediately preceding pixel is compared. Up to the pixels are extracted as signal flat portions. When the signal flat portion data is recorded, the area on the plane is masked as a mask area.

  Here, the state of the signal flat portion processing will be described with reference to FIG. For example, if the image is composed of 720 pixels wide × 480 pixels long, the processing is sequentially performed from the upper left pixel (1, 1) to the lower right pixel (720, 480). 1,1) to the right uppermost pixel (720,1), and then processing in the right direction, then the rightmost second row pixel (720,2) to the leftmost second row pixel (1,2) leftward Further, the processing is performed in a zigzag manner from the pixel (1, 3) in the third row on the left end to the pixel (720, 3) in the third row on the right end, and the right end 479. The processing is performed from the pixel (720, 479) on the row through the pixel (720, 480) on the bottom rightmost row to the pixel (1, 480) on the bottom leftmost row. When the values of consecutive pixels are the same in this processing direction, they are recorded as signal flat portion data. The reason for reversing the odd-numbered and even-numbered processing directions in this way is to suppress fluctuations in pixel values to be processed. For example, in the example of FIG. 3, when attention is paid to the pixel value of the pixel (720, 1), it is expected that the pixel value of the pixel (720, 2) is not significantly different from the pixel value of the pixel (720, 2). Is expected to change significantly. In the processing of step S2 and step S3, which will be described later, data compression is possible when the difference between pixel values is “0”. Therefore, the odd-numbered and even-numbered processing directions are reversed in order to suppress variations in pixel values. It is.

  The signal flat portion data is created by recording the coordinate value of the first pixel of the signal flat portion and recording the number of pixels and pixel values having the same pixel value in succession in the signal flat portion table. An example of the flat signal portion table is shown in FIG. In the present embodiment, the signal flat portion data is composed of the ij coordinate of the first pixel, the number of consecutive pixels, and a pixel value common to the signal flat portion. The signal flat part shown in the first row in FIG. 4A indicates that the values of (5, 10) to 85 (that is, (5, 94)) pixels are “136” and are the same. . The signal flat portion shown in the second row of FIG. 4 indicates that the values of 106 pixels from (681, 25) are “88” and are the same. When the number is 106 from (681, 25), it reaches (720, 25) at the right end, and (720, 26) to (655, 26) in the lower row are signal flat portions.

  Note that the signal flat part table shown in FIG. 4A is prepared for each frame number and plane type, so that it can be understood to which plane of which frame the signal flat part data should be returned at the time of decoding. It has become.

  Next, a computation process between temporally adjacent frames is performed on the digital video signal subjected to the signal flattening process (step S3). Specifically, first, a difference calculation is performed between pixels having the same coordinates on the same color planes between adjacent frames. As a result, a set of pixels having a difference of “0”, that is, having the same pixel value is separately recorded as the same data between frames. The pixel difference calculation between adjacent frames in step S3 is also performed in the pixel scanning line direction, similar to step S2. Specifically, as long as the difference between adjacent frames with the same coordinate pixel is “0”, the next pixel is also compared, and there is a section where the difference is “1” or more. The pixels up to the pixel just before are extracted as the same data between frames. At this time, the region already masked by the signal flat portion process is excluded from the comparison target pixels. When the same data is recorded between frames, the area of the subsequent frame is masked as a mask area. In addition, between the adjacent frames is, for example, the frame F1 and the frame F2 in the example of FIG. 1B, and the frame F2 is a subsequent frame. When the correlation calculation is performed between the frames F1 and F2, 3 of the X plane of the frame F1 and the X plane of the frame F2, the Y plane of the frame F1 and the Y plane of the frame F2, the Z plane of the frame F1 and the Z plane of the frame F2 Two correlation operations are performed.

  Difference calculation between frames is performed by comparing pixels having the same coordinates on the same plane. That is, after comparing the pixels (1, 1), the pixels (2, 1) are compared, and the processing direction is reversed for each row, as in the case of the signal flat portion shown in FIG. Processing is performed up to (1,480). As a result, the section where the difference between the pixels is “0” is masked as a mask area and recorded as the same data between frames.

  To create the same data between frames, the coordinate value of the first pixel with a difference of “0” is recorded, and the number of consecutive pixels with a difference of “0” is recorded in the table between frames. Done. An example of the interframe table is shown in FIG. In this embodiment, the same data between frames is composed of the ij coordinate of the first pixel and the number of continuous pixels.

  Note that the inter-frame table shown in FIG. 4B is prepared for each inter-frame and for each plane type, so that it can be seen which plane of which frame the same inter-frame data should be returned at the time of decoding. It has become.

  As described above, when the pixel difference calculation is performed between both adjacent frames, the difference in pixel value naturally increases in a portion where the movement is fast. For example, in a scene where an object is moving violently at a specific position in a certain landscape, it is known in advance that the correlation is low at the position where the object exists, and therefore, only the correlation calculation process is wasted. In such a case, it is only necessary to set only a fast-moving part as a non-application area for the inter-frame calculation and perform the inter-frame calculation only for the other areas. Therefore, in the present invention, the frame of the image is displayed on the display means or the like, and the position of the non-application area on the frame and the frame number (time) are set in advance by the user. As described above, for the frame in which the non-application area is set, the correlation calculation is performed on the area excluding the non-application area.

  As described above, the signal flat portion data and the same data between frames are separated from the frame group of the digital video signal, and the pixels of each separated data are recorded in each table with a small amount of data in a so-called run length format. The On each plane, areas corresponding to pixels extracted as signal flat portion data and identical data between frames are masked as mask areas.

  The mask area is performed by recording the coordinates of the masked section in each plane of each frame in the mask data. In the processing for a non-mask area to be described later, the computer recognizes the mask area by referring to the mask data, and recognizes the area other than the mask area as the non-mask area, thereby determining the pixel to be processed.

  Subsequently, for the pixels in the non-mask area, the prediction value and the prediction error of each pixel are calculated based on the neighboring pixels (step S4). In the present embodiment, in order to calculate the prediction error value e (i, j) of the target pixel (i, j) to be calculated, the original pixel value x (i, j) of the target pixel and FIG. The values x (i−1, j−1) and x (i, j−) of three neighboring pixels (i−1, j−1), (i, j−1), (i−1, j) as shown in FIG. 1), x (i-1, j) is used.

Hereinafter, prediction error conversion using such three neighboring pixels will be described. A processing outline of the prediction error conversion is shown in the flowchart of FIG. First, a linear prediction error corresponding to each prediction calculation formula is calculated using a plurality of prediction calculation formulas prepared in advance (step S11). Specifically, the following [Formula 2] to [Formula 8] are prepared as prediction calculation formulas for calculating the prediction error of the pixel (i, j).
[Formula 2]
e0 = x (i, j) -x (i-1, j)
[Formula 3]
e1 = x (i, j) -x (i, j-1)
[Formula 4]
e2 = x (i, j) -x (i-1, j-1)
[Formula 5]
e3 = x (i, j) -x (i-1, j) -x (i, j-1) + x (i-1, j-1)
[Formula 6]
e4 = x (i, j) -x (i-1, j)-{x (i, j-1) -x (i-1, j-1)} / 2
[Formula 7]
e5 = x (i, j) -x (i, j-1)-{x (i-1, j) -x (i-1, j-1)} / 2
[Formula 8]
e6 = x (i, j)-{x (i-1, j) + x (i, j-1)} / 2
In the above [Equation 2] to [Equation 8], e0 to e6 are prediction errors of the pixel (i, j) by each prediction calculation equation, and x (i, j), x (i, j-1), x (I-1, j) and x (i-1, j-1) are pixels (i, j), (i, j-1), (i-1, j), (i-1, j-1). ).

  Subsequently, the linear prediction error that minimizes the cumulative error, which is the cumulative absolute value of the prediction error value for each prediction calculation formula, is selected as the prediction error of the pixel (step S12). Here, the concept of cumulative error is used. Specifically, the cumulative values of past pixels of the prediction error calculated by the respective prediction calculation formulas [Formula 2] to [Formula 8] are set as A0 to A6. Then, a prediction error corresponding to the smallest one of the accumulated errors A0 to A6 is selected. For example, it is assumed that A3 is the smallest among A0 to A6. In this case, the prediction error e3 calculated by [Formula 5] is selected as the prediction error e (i, j) to be encoded. The selected prediction error e (i, j) is replaced with the original value x (i, j) of the pixel (i, j), and the subsequent processing is performed.

Subsequently, the absolute values of the prediction errors e0 to e6 are added to the accumulated errors A0 to A6 (step S13). Specifically, as shown in [Formula 9] below, the variables A0 to A6 that are accumulated error values are updated. At the same time, each time the processing of each sample is performed, a process of incrementing the counter by one is performed.
[Formula 9]
A0 ← A0 + | e0 | A1 ← A1 + | e1 |
A2 ← A2 + | e2 | A3 ← A3 + | e3 |
A4 ← A4 + | e4 | A5 ← A5 + | e5 |
A6 ← A6 + | e6 |
Subsequently, it is determined whether the counter has exceeded a predetermined number of times (step S14). In this embodiment, this predetermined number is set as 100 times. That is, it is determined whether or not the counter exceeds 100.

As a result, if the counter exceeds 100, the accumulated error is halved (step S15). Specifically, as shown in the following [Formula 10], variables A0 to A6, which are accumulated errors, are divided by two. At the same time, the counter is reset to zero. That is, A0 to A6 here are not cumulative errors in a pure sense, but are moving averages of cumulative errors. In the present embodiment, processing is performed so that the previous maximum 100 pixels are accumulated, but the previous one is halved. As a result, the influence of pixels that are distant from each other is reduced.
[Formula 10]
A0 ← (A0) / 2 A1 ← (A1) / 2
A2 ← (A2) / 2 A3 ← (A3) / 2
A4 ← (A4) / 2 A5 ← (A5) / 2
A6 ← (A6) / 2
By executing the processing of step S11 to step S15 for all pixels in the non-mask area of all frames and all planes, the value of all pixels in the non-mask area changes from the original value x (i, j) to the target error. It will be replaced by e (i, j). In each pixel data on the plane, there is a mask area masked in the signal flat portion processing in step S2 and the inter-frame calculation in step S3. Each pixel on the mask area is not a target pixel for prediction error calculation, but is used to calculate a prediction error for a target pixel that does not belong to the mask area. For example, in a plane as shown in FIG. 5, when obtaining a prediction error using a pixel (i, j) that does not belong to the mask area as a target pixel, the pixel (i−1, j−1), the pixel (i, j−1) ), Pixel (i-1, j), pixel (i-1, j-1), pixel (i, j-1), pixel (i- 1, j) is used to calculate the prediction error of pixel (i, j).

  When predictive encoding is performed as described above, the positive / negative processing of each pixel data is subsequently performed (step S5). Although the value of each pixel has been replaced by the prediction error of the X, Y, and Z values from the R, G, and B values by the processing from step S1 to step S4, the bit format of each pixel remains unchanged. is there. Normally, when being calculated by a computer such as a computer, each data is processed in units of 32 bits and expressed using a two's complement expression. This is converted into a positive / negative signed absolute value expression, and the absolute value portion is moved one bit higher, and the positive / negative sign bit is moved to LSB (least significant bit). FIG. 7 schematically shows how the bit structure is converted in step S5. FIG. 7A shows a bit configuration before processing, and FIG. 7B shows a bit configuration after processing. The reason why the positive and negative code bits are moved to the LSB in this way is to make it easier to detect the bit length of each pixel in the variable length encoding process in the subsequent step S6.

  Next, a process of converting each pixel to a variable length is performed (step S6). The variable length coding in this embodiment employs a method generally called Golomb coding. Specifically, the bit component constituting one pixel is divided into an upper bit component and a lower bit component, the lower bit component is left unchanged, and the upper bit component is a numerical value obtained by decimal conversion of only the upper bit. It is an array in which bit “0” is arranged and separator bit “1” is added at the end. For example, consider an 8-bit bit component “00101000”. At this time, if the lower bit component is 4 bits, the lower bit component is “1000”. Since the upper bits are “0010”, “2” pieces of “0” obtained by decimal conversion are arranged and converted to “001” by adding “1” at the end. As a result, the 8-bit bit string “00101000” is converted into a 7-bit bit string “0011000”. In the present embodiment, the bit length of the lower-order bit component that makes the bit component unchanged before and after conversion is made variable in each pixel.

  Hereinafter, the variable length encoding process in step S6 will be specifically described. FIG. 8 is a flowchart showing an outline of variable length coding. First, an average bit length Bf, which is a moving average of the bit lengths of processed pixels, is calculated (step S21). The average bit length Bf is obtained by dividing the cumulative bit length RB, which is the cumulative value of the processed bit length, by a counter C based on the number of processed pixels. That is, Bf = RB / C. Since the accumulated bit length RB is 0 in the initial state, when the pixel of (i, j) = (1, 1) is processed, the bit length of the pixel of (i, j) = (1, 1) Bd (i, j) is set as an initial value. The initial counter C = 1 is set.

  Subsequently, the bit length Bd (i, j) of the pixel at the coordinates (i, j) is calculated (step S22). For pixels after (i, j) = (2, 1), after calculating the average bit length Bf, the bit length Bd (i, j) of the pixel is calculated. This bit length Bd (i, j) is easily calculated by converting the bit configuration in step S5. By converting to the bit configuration as shown in FIG. 7B, the bit length starts from the point where the bit “1” appears at the head in the bit configuration of each pixel. Next, the bit length Bv of the changing unit is calculated (step S23). This is calculated by subtracting the average bit length Bf from the bit length Bd (i, j) of the pixel. Subsequently, the data code is output (step S24). Specifically, “0” is output for the numerical value obtained by decimal conversion of the upper Bv bits, then the separator bit “1” is output, and the lower Bf bits are output as the unchanged part. The code output is performed as recording to an external storage device such as a hard disk or a CD-R. Next, the bit length Bd (t) is added to the cumulative bit length RB (step S25). At the same time, every time processing of each pixel is performed, processing of incrementing the counter C one by one is performed. Subsequently, it is determined whether or not the counter C exceeds a predetermined number (step S26). Here, about 100 is set as the predetermined number. Therefore, it is determined whether or not the counter exceeds 100. As a result, if the counter exceeds 100, the cumulative bit length RB is halved (step S27). Specifically, the variable RB that is the cumulative bit length is divided by two. At the same time, the counter C is halved.

As described above, encoding with variable bit length is performed for each pixel. A set of pixels expressed by variable bit length obtained by encoding is output as variable length prediction error data.
(Recording of code data)
The code data obtained as described above is recorded in units of frames. Here, the state of the code data recorded in units of frames is shown in FIG. That is, for each frame, the signal flat portion data, the same data between frames, and the variable length prediction error data are recorded. These data are recorded in a format suitable for the recording medium to be recorded.
(Decryption method)
Next, a method for decoding and decompressing code data compressed by the compression method will be described. Decoding is executed by a computer and a dedicated software program installed in the computer. An outline of the decoding method is shown in the flowchart of FIG.

  First, a recording medium on which code data as shown in FIG. 9 is recorded is read by an apparatus for decoding (computer equipped with dedicated software). Reading is sequentially performed in units of frames. First, the signal flat part data is read, the pixel values of the signal flat part are arranged in each plane, mask data is created, and the pixels in the signal flat part region are masked (step S31). Subsequently, the same data between frames is read, and the corresponding pixel value is arranged in the previous frame for the corresponding area of each plane, and the arranged pixel area is masked in the mask data created in step S31. (Step S32). Note that the process in step S32 is not performed on the first frame. This is because there is no identical data between corresponding frames in the first frame. Next, the variable length prediction error data is read, and the variable length prediction error data is restored to data in which each pixel is expressed in a fixed length (step S33). This may be the reverse of the process in step S6. Subsequently, the polarity of the data restored for each pixel is restored (step S34). This may be the reverse of the process in step S5. Next, the data of each pixel whose polarity has been restored is arranged on each plane with reference to the mask data (step S35). Specifically, the pixel data is arranged at the coordinates corresponding to the non-mask area other than the mask area recorded in the mask data as in the encoding. That is, pixels (1, 1) to pixels (1, 480) are sequentially arranged in the direction of the arrow shown in FIG.

  At this time, since the value of each pixel at each coordinate (i, j) is recorded with the prediction error e (i, j), it is restored to the original pixel value x (i, j) (step S36). . This may be the reverse of the process in step S4.

  At this point, the frame group is configured with R, G, and B planes because each frame is configured with X, Y, and Z planes as shown in FIG. The original frame is restored (step S37). This can be restored by using the above [Equation 1]. By repeating the processing from step S31 to step S37 for all frames, the digital video signal is restored. Since the compression method described in the present embodiment is a lossless type (reversible type), the restored video signal is not lost and is exactly the same as the video signal before compression.

  In the above embodiment, the case where an RGB color video signal is input as an original digital video signal to be compressed has been described as an example. However, it is also possible to input a monochrome video signal as an original digital video signal and perform compression. is there. In this case, since the pixel values of the X and Z planes are all “0” in the above [Equation 1], only the process for the Y plane needs to be performed.

The preferred embodiment of the present invention has been described above. However, the present invention is not limited to the above embodiment, and various modifications can be made. Hereinafter, modified examples will be described.
(Lossy type)
In the above-described embodiment, the compression is performed by the lossless type that can be completely restored to the original state. However, it is also possible to adopt a near lossless type that is irreversible but obtains an image quality close to lossless compression. In this case, the arithmetic processing between the color signals in step S1 is different. Specifically, as shown in FIG. 1, instead of converting R, G, and B signals into X, Y, and Z, the following [Formula 11] is used to obtain Y ₂ , U, V, Convert to
[Formula 11]
Y ₂ = 0.299R + 0.587G + 0.114B
U = -0.1684R-0.3316G + 0.5B
V = 0.5R-0.4187G-0.0813B
Since the above [Formula 11] involves a decimal point operation, the original values of R, G, and B cannot be restored from Y ₂ , U, and V converted to integer values. Become a mold. Further, for the U plane and the V plane, the pixels are thinned out in half in the vertical direction and the horizontal direction. For example, in the case of an image of 720 × 480 pixels, an image of 360 × 240 pixels is obtained, and the number of pixels is reduced to ¼. FIG. 11 shows the state of conversion from R, G, B signals to Y ₂ , U, V signals. FIG. 11A is exactly the same as FIG. 1A and shows an input digital video signal. As shown in FIG. 11B, the U plane and the V plane have half the number of pixels in the vertical and horizontal directions. The Y ₂ , U, and V planes thus obtained are subjected to the processing from step S 3 to step S 6, and irreversible compression is performed. Although there is almost no loss for the Y ₂ plane, The resolution of the V plane is reduced to ¼. However, since the human visual system is known to be insensitive to the color difference signals U and V planes compared to the luminance signal Y ₂ plane, it can recognize the deterioration of image quality as a whole. It is difficult to obtain an image quality almost equivalent to a reversibly compressed image (referred to as near lossless coding).

  As described above, since the present invention is performed on a video signal composed of a plurality of still images, when the input data is a still image, it can also be used as a still image compression method. In this case, the still image can be compressed by executing steps S1 to S6 other than step S3.

It is a schematic diagram which shows the structure of the digital video signal made into compression object by the compression method of the video signal of this invention. It is a flowchart which shows the outline | summary of the compression method of the video signal which concerns on this invention. It is a figure which shows the process order of the pixel on a flame | frame. It is a figure which shows the mode of a signal flat part table and an inter-frame table. It is a figure which shows the positional relationship of each pixel at the time of performing a prediction error calculation process. It is a flowchart which shows the detail of prediction error calculation. It is a figure which shows the mode of the process of a polarity process. It is a flowchart which shows the outline | summary of variable length coding. It is a figure which shows the code data obtained by the compression method of the video signal which concerns on this invention. It is a flowchart which shows the outline | summary of the decoding process of the compressed video signal. It is a schematic diagram which shows the mode of the calculation between color signals in the case of performing lossy compression.

Claims (10)

A method of compressing an information amount so that a video signal configured as a frame group in which frames that are still images are arranged in time series can be reproduced,
A signal flat part for detecting a section in which each pixel data on each frame is the same in accordance with the scanning line direction, masking the detected section as a mask area, and extracting information of the detected section as signal flat part data Processing stage;
Using each pixel existing in a non-masked area that has not been masked in each frame as a target pixel, and using the pixel value of the target pixel and the pixel value of a neighboring pixel located in the vicinity of the target pixel, the target pixel And calculating a plurality of prediction error candidates for a plurality of prediction calculation formulas having different linear coefficients to be multiplied by pixel values of the neighboring pixels, and adopting an optimum one of the prediction error candidates as a prediction error of the target pixel A prediction error calculating step of replacing the value of the target pixel with the prediction error;
A variable length encoding step of encoding each pixel data replaced with the prediction error with a variable length;
A method for compressing a video signal, comprising: In claim 1,
A method of compressing video signals, wherein the processing direction of each pixel data on each frame in the signal flat part processing step is performed in the opposite direction between the odd-numbered scanning lines and the even-numbered scanning lines. In claim 1,
When the video signal is composed of the three primary color components of RGB, the calculation between the color signals is converted into the XYZ component by the conversion formula of X = RG, Z = BG, Y = G + (X + Z) / 3 The step is provided immediately before the signal flat portion processing step, and the signal flat portion processing step, the prediction error calculation step, and the variable length coding step are processed separately for X, Y, and Z components. A compression method of the video signal. In claim 1,
When the video signal is composed of RGB three primary color components, the video signal is converted into a luminance signal component Y ₂ , a first color signal component U, and a second color signal component V, and the first color signal component U and the first color signal component U are converted. An inter-color signal calculation step for thinning out pixels at a predetermined ratio for the two-color signal component V is provided immediately before the signal flat portion processing step, and the signal flat portion processing step, the prediction error calculation step, A method of compressing a video signal, wherein the variable length coding step is processed for each of Y ₂ , U, and V components. In claim 1,
For each frame constituting the frame group, a frame in which a pixel having the same value as a pixel at the same position in the immediately preceding frame is masked as a mask region, and information on the interval is extracted as the same data between frames. A method for compressing a video signal, characterized in that an intermediate calculation step is provided before the prediction error calculation step. In claim 1,
The neighboring pixels in the prediction error calculation stage are included in the eight neighboring ranges of the target pixel, and are three pixels positioned forward in the scanning line process, and each target positioned forward in the scanning line process Accumulated absolute values of the prediction error candidates calculated in the pixel are calculated for each prediction calculation formula, and the prediction error candidate calculated by the prediction calculation formula that minimizes the cumulative value is used as the prediction error of the target pixel. A video signal compression method characterized by being adopted. In claim 1,
A video signal characterized by having a polarity processing step that shifts the whole to the upper one bit with respect to the absolute value of the prediction error value and inserts positive / negative polarity bits into the least significant bit before the variable length encoding step. Compression method. In claim 1,
The variable length encoding step encodes the lower bit component of the bit component of each pixel converted into the prediction error value with the bit component as it is, and changes the bit component with respect to the remaining upper bit component. And a video signal compression method characterized in that encoding is performed. A method of compressing the amount of information for a still image that is a set of pixels so that the still image can be reproduced,
A signal flat part that detects a section in which each pixel data on the still image is the same in accordance with a scanning line direction, masks the detected section as a mask area, and extracts information of the detected section as signal flat part data Processing stage;
Using each pixel existing in a non-masked area that has not been masked in each frame as a target pixel, and using the pixel value of the target pixel and the pixel value of a neighboring pixel located in the vicinity of the target pixel, the target pixel And calculating a plurality of prediction error candidates for a plurality of prediction calculation formulas having different linear coefficients to be multiplied by pixel values of the neighboring pixels, and adopting an optimum one of the prediction error candidates as a prediction error of the target pixel A prediction error calculating step of replacing the value of the target pixel with the prediction error;
A variable length encoding step of encoding each pixel data replaced with the prediction error with a variable length;
A method of compressing a still image, comprising: The recording medium which recorded the code | cord | chord data output by the method in any one of Claims 1-9 with respect to the said video signal.

Images