JP5748225B2 - Moving picture coding method, moving picture coding apparatus, and moving picture coding program - Google Patents

Description

  The present invention relates to a moving picture coding method, a moving picture coding apparatus, and a moving picture coding program for improving the selection accuracy of skip macroblocks in video coding.

  Conventionally, there is a technique described in Patent Document 1 for the purpose of accurately detecting a small block belonging to a skip area that does not have information that needs to be encoded. In this prior art, for each small block composed of a plurality of pixels, a total prediction error with respect to the previous frame is obtained and compared with a first threshold, and a variance within the small block is obtained to obtain a second threshold and In comparison, when the total prediction error with the previous frame is equal to or smaller than the first threshold and the variance is equal to or smaller than the second threshold, it is determined that the small block belongs to the background area. The skip mode (the amount of generated code is the smallest and the amount of computation is the smallest in the inter prediction mode) is applied to the background area without motion determined in this way to reduce the amount of code and speed up the computation. ing.

  FIG. 12 shows a block diagram of a conventional apparatus for detecting a skip area. The prediction error evaluation unit 30 obtains the total prediction error of the encoded block of the previous frame stored in the encoded image memory 34 for each encoding target macroblock, and compares it with the first threshold value. The variance in the macroblock is obtained and compared with a second threshold, and the macroblock belongs to the background area when the total prediction error is less than or equal to the first threshold and the variance is less than or equal to the second threshold. Is determined. Otherwise, it is determined that the macroblock is not a skip area, and skip mode OFF information is notified to the encoding unit 32.

  When it is determined that the macroblock belongs to the background area, the prediction vector evaluation unit 31 determines whether the prediction motion vector (or the motion vector of the encoding target macroblock) stored in the motion vector memory 33 is 0, If it is not 0, the encoding unit 32 is notified of skip mode OFF information. If the predicted motion vector is 0, the encoding target macroblock is determined to be a skip area, and skip mode ON information is notified to the encoding unit 32. As the predicted motion vector, for example, a median of motion vectors around the encoding target macroblock is often used.

  The encoding unit 32 encodes the macro block to which the skip mode ON information is added in the skip mode, and the macro block to which the skip mode OFF information is added is encoded using normal inter-screen prediction other than the skip mode. I do. The motion vector resulting from the encoding is stored in the motion vector memory 33 for reference in subsequent predictive encoding, and the decoded image signal of the encoded block is stored in the encoded image memory 34.

JP 2002-281508 A

In the prior art disclosed in Patent Document 1,
・ Procedure 1: Calculate the prediction error between the encoding target macroblock and the encoded block corresponding to the true block,
Procedure 2: If the prediction error is smaller than the threshold value, it is regarded as a background area, and the skip mode is applied.

In this conventional technique, the prediction error of the encoding target frame and the encoded reference frame is used for the evaluation. Therefore, if the prediction error is smaller than the threshold value, it is highly possible that the frame is stationary and the subjective image quality is not impaired. It can be determined that the skip mode can be applied. However, since the encoded block is actually distorted by quantization, in the case of a low bit rate (quantization width is large),
-Although it is subjectively stationary, the prediction error increases due to coding distortion.
・ Prediction error has increased due to movement,
As a result, it is difficult to identify either of them by the threshold value, and as a result, there is a problem that the subjective image quality is significantly impaired by applying the wrong skip mode, or that the area where the skip mode can be applied is missed. is there. That is, there are the following problems.

(1) Loss of subjective image quality and skipped area skipping due to threshold setting If the motion area is mistakenly detected as the background area and the skip mode is selected, the subjective image quality will be distorted and the subjective image quality will be greatly increased. It will be damaged. In order to avoid this, the problem can be solved by setting the threshold value in the procedure 2 low. However, on the other hand, since the error tends to increase due to the influence of coding noise in the background area, it is determined that the skip macroblock is not applied when the error is higher than the threshold value. I'm missing an area.

  (2) The skip area cannot be appropriately applied to both the background block with high variance and the background block with low variance.

  In general, even in the background region, the size of the prediction error varies depending on the texture. The background error with low variance tends to have a low prediction error, and the block with high variance tends to have a high prediction error. Therefore, with a single threshold, neither a highly dispersed background block nor a low background block can be detected properly at the same time. That is, if the threshold is set low, a highly dispersed background block is missed. If the threshold value is set high, it is possible to cover even highly dispersed background blocks, but there is a possibility that the motion area is erroneously detected as a skip application area.

  Therefore, skip mode is applied to general video that solves the above problems, regardless of the level of dispersion, and blocks that are visually considered to be background are not leaked. An algorithm that performs motion search is needed.

  Furthermore, since the skip mode is used for speeding up the codec in addition to reducing the code amount, the computation load of the skip mode application area extraction algorithm itself must be low.

  An object of the present invention is to solve the above-described problems, improve the accuracy of skip macroblock selection over the prior art, improve subjective image quality under the same generated code amount conditions, and reduce the amount of computation. .

  In order to solve the above problems, the present invention selects and encodes skip macroblocks by the following method.

[Method 1]
(1) First, when the encoding target macroblock is a bi-predictive image, the total of the block pixel values of the encoding target macroblock and the original image corresponding to the first encoded reference frame in the temporal / spatial direct mode The difference macro value ΔDC with respect to the sum of the block pixel values of the macroblock at the same position (true back) as the encoding target macroblock in the frame or the encoding target macroblock in the original frame corresponding to the second encoded reference frame The larger difference absolute value ΔDC is calculated from the difference absolute value ΔDC with respect to the total of the block pixel values of the macroblocks at the same position . The difference absolute value of the average pixel value of each block may be ΔDC.

(2) When the difference absolute value ΔDC is larger than the predetermined threshold Th1, the skip mode is not selected. When the difference absolute value ΔDC is equal to or smaller than the threshold value Th1 , a prediction error (SAD: Sum of Absolute Difference) or second between the macro block to be encoded and the reference block in the first encoded reference frame in the temporal / spatial direct mode Among the prediction errors with the reference block in the encoded reference frame, the larger prediction error is calculated. Other prediction errors such as SSD (Sum of Square Difference) may be used.

  (3) If the obtained prediction error is smaller than the predetermined threshold Th2 and the prediction motion vector (PMV) or the motion vector of the encoding target macroblock is zero, the skip mode is selected. Otherwise, Predictive coding is performed without using the skip mode. Note that the predicted motion vector or the motion vector may not be completely zero, and may be regarded as zero if it is equal to or less than a predetermined value.

[Operation 1]
The operation of Method 1 is as follows.

In the invention of the method 1, a skip application area is determined by newly introducing an error of an original picture frame corresponding to an encoding target frame and an encoded reference frame and combining it with a prediction error. Since the original image frame is not subjected to quantization distortion, the prediction error of the block that is subjectively stationary becomes small and the movement region can be easily distinguished, except for the influence of noise from the camera. Furthermore, in the present technology, since the error of the original image is evaluated using ΔDC corresponding to the error of the average pixel value of the block, it is possible to extract a still region having a high noise component.
However,
・ A small movement area that cannot be determined as movement only with ΔDC,
-Even if ΔDC is small, the region where the prediction error is large due to the effect of coding distortion,
Therefore, these regions can be excluded by evaluating the SAD of the prediction error. By combining both evaluation methods, the false detection of the skip mode area is reduced compared to the conventional technique, so that the subjective image quality can be improved and the skip mode can be prevented from being missed. It is possible to suppress the calculation amount of the entire codec (even if the error calculation amount is included). In addition, since the unnecessary SAD calculation can be avoided by performing the first evaluation using ΔDC, it is possible to reduce the calculation of the skip mode selection algorithm itself.

[Method 2]
The present invention can also effectively solve the above problems by the following method.

(1) First, when the encoding target macroblock is a bi-predictive image, the total of the block pixel values of the encoding target macroblock and the original image corresponding to the first encoded reference frame in the temporal / spatial direct mode The difference macro value ΔDC with respect to the sum of the block pixel values of the macroblock at the same position (true back) as the encoding target macroblock in the frame or the encoding target macroblock in the original frame corresponding to the second encoded reference frame The larger difference absolute value ΔDC is calculated from the difference absolute value ΔDC with respect to the total of the block pixel values of the macroblocks at the same position . The difference absolute value of the average pixel value of each block may be ΔDC.

(2) When the difference absolute value ΔDC is larger than the predetermined threshold Th1, the skip mode is not selected. When the difference absolute value ΔDC is equal to or smaller than the threshold value Th1 , a prediction error (SAD: Sum of Absolute Difference) or second between the macro block to be encoded and the reference block in the first encoded reference frame in the temporal / spatial direct mode Among the prediction errors with the reference block in the encoded reference frame, the larger prediction error is calculated. Other prediction errors such as SSD (Sum of Square Difference) may be used.

  (3) Calculate the variance of the pixel values of the encoding target macroblock, and select a large value as the threshold Th2 if the variance is large, and select a small value as the threshold Th2 if the variance is small. Set. Here, the distribution is not limited to the distribution in a narrow sense, but may be a distribution in a broad sense such as a so-called activity. The activity is, for example, the maximum value of four values obtained by determining the distribution of individual small blocks when the block is divided into four.

  (4) If the prediction error obtained in (2) is smaller than the predetermined threshold Th2 and the predicted motion vector (PMV) or motion vector of the encoding target macroblock is zero, the skip mode is selected, and so on. Otherwise, predictive coding is performed without using the skip mode. Note that the predicted motion vector or the motion vector may not be completely zero, and may be regarded as zero if it is equal to or less than a predetermined value.

[Action 2]
The operation of Method 2 is as follows.

  When the error (prediction error) between the encoding target frame and the encoded reference frame is actually calculated, the prediction error is large and the variance is low in a block with high variance even in a subjectively stationary region The prediction error is smaller in the block.

  In order to take this phenomenon into account, in the technique of method 2, the variance of the encoding target frame is calculated in advance, and a high threshold is set for a block with a large variance and a low threshold is set for a block with a small variance. By setting a low threshold for a block with small variance, it is possible to exclude a block with a small prediction error from the skip mode target despite the occurrence of motion (avoiding false detection).

  In addition, by setting a high threshold for blocks with high variance, blocks with a large prediction error even though they are subjectively stationary can be extracted as skip mode targets. Reduction). Therefore, since the erroneous detection and missing of the skip mode target area can be prevented more than the method 1 invention, it is possible to further improve the subjective image quality and reduce the calculation amount of the entire codec.

  Furthermore, the present invention can be similarly applied to bi-predictive pictures, and can be similarly applied to interlaced images.

  Compared with the prior art, the present invention makes it possible to improve the subjective image quality under the same code amount condition and to reduce the calculation amount of the encoder / decoder with the improvement of skip macroblock selection accuracy. In addition, it is possible to reduce the amount of calculation required for skip macroblock selection area extraction.

It is a figure which shows the example of 1 structure of the moving image encoder with which this invention is applied. It is a figure which shows the structural example of Example 1 of the part of the skip mode control part. 3 is a flowchart of the first embodiment. It is a figure explaining the difference between Example 1 and Example 2. FIG. It is a figure which shows the structural example of Example 2 of the part of the skip mode control part. 10 is a flowchart of Example 2. FIG. 10 is a diagram for explaining Example 3; 10 is a flowchart of Example 3. 10 is a flowchart of Example 4. FIG. 10 is a diagram for explaining a fifth embodiment. It is a figure which shows the structural example of the system in the case of implement | achieving a moving image encoder using a computer and a software program. It is a block diagram of the apparatus which detects the conventional skip area | region.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 shows an example of the configuration of a moving picture encoding apparatus to which the present invention is applied. In the moving image encoding apparatus 100, in the present embodiment, in particular, the skip mode control unit 1021 in the inter prediction processing unit 102 is different from the related art, and the other part is H.264. The configuration is the same as that of a conventional general video encoding apparatus used as an encoder such as H.264 / AVC.

  The video encoding apparatus 100 receives a video signal to be encoded, divides a frame of the input video signal into blocks, encodes each block, and outputs the bit stream as an encoded stream. For this encoding, the prediction residual signal generation unit 103 obtains a difference between the input video signal and the prediction signal output from the intra prediction processing unit 101 or the inter prediction processing unit 102, and outputs the difference as a prediction residual signal. To do. The transform processing unit 104 performs orthogonal transform such as discrete cosine transform (DCT) on the prediction residual signal and outputs a transform coefficient. The quantization processing unit 105 quantizes the transform coefficient and outputs the quantized transform coefficient. The entropy encoding processing unit 111 performs entropy encoding on the quantized transform coefficient and outputs it as an encoded stream.

  On the other hand, the quantized transform coefficient is also input to the inverse quantization processing unit 106 where it is inversely quantized. The inverse transform processing unit 107 performs inverse orthogonal transform on the transform coefficient output from the inverse quantization processing unit 106, and outputs a prediction residual decoded signal.

  The decoded signal generation unit 108 adds the prediction residual decoded signal and the prediction signal output from the intra prediction processing unit 101 or the inter prediction processing unit 102 to generate a decoded signal of the encoded block to be encoded. This decoded signal is stored in the frame memory 109 for use as a reference image in the intra prediction processing unit 101 or the inter prediction processing unit 102. In addition, when referring in the inter prediction process part 102, in the in-loop filter process part 110, the image stored in the frame memory 109 is input, the filtering process which reduces an encoding distortion is performed, and after this filtering process Use the image as a reference image.

  Information such as the prediction mode set in the intra prediction processing unit 101 is entropy encoded in the entropy encoding processing unit 111 and output as an encoded stream.

  Information such as a motion vector set in the inter prediction processing unit 102 is also entropy encoded in the entropy encoding processing unit 111 and output as an encoded stream.

  The skip mode control unit 1021 determines whether the encoding target block has information that needs to be encoded, and the generated code amount is minimum for the block determined not to be encoded, and Control is performed to apply the skip mode with the smallest amount of computation in the inter-screen prediction mode.

  Hereinafter, a method for realizing the skip mode control unit 1021 in the present embodiment will be described in detail.

[Example 1]
FIG. 2 shows a configuration example of the first embodiment of the skip mode control unit 1021. In FIG. 2, the part from the macroblock error calculation unit 10 to the prediction vector evaluation unit 13 mainly corresponds to the skip mode control unit 1021 in FIG. 1, and the encoding unit 14 is an inter prediction processing unit other than the skip mode control unit 1021. 102 and the entropy encoding processing unit 111 and the like. The encoded image memory 15 and the motion vector memory 16 correspond to the frame memory 109 and the like.

Macroblock error calculator 10, the encoded macroblock, a difference absolute sum of the block pixel values of the macroblock coded macroblock at the same position in the original frame corresponding to the encoded reference frame (right in back) The value ΔDC is calculated.

  The macroblock error evaluation unit 11 compares whether or not the absolute difference value ΔDC is greater than a predetermined threshold value Th1, and if the absolute difference value ΔDC is greater than the predetermined threshold value Th1, notifies the encoding unit 14 of skip mode OFF information. . When the difference absolute value ΔDC is equal to or smaller than the threshold Th1, the prediction error evaluation unit 12 is notified of the evaluation result to that effect.

  The prediction error evaluation unit 12 calculates a prediction error (SAD) between the encoding target macroblock and the reference encoded block of the encoded reference frame stored in the encoded image memory 15, and calculates a predetermined threshold Th2. Compare with When the prediction error is larger than the threshold value Th2, skip mode OFF information is notified to the encoding unit 14. When the prediction error is equal to or smaller than the threshold Th2, the evaluation result to that effect is notified to the prediction vector evaluation unit 13.

  The predicted vector evaluation unit 13 determines whether the predicted motion vector (PMV) stored in the motion vector memory 16 is 0, and if not, notifies the encoding unit 14 of skip mode OFF information. If the motion vector predictor is 0, the encoding target macroblock is determined to be a skip area, and skip mode ON information is notified to the encoding unit 14.

  The encoding unit 14 encodes the macro block to which the skip mode ON information is added in the skip mode, and the macro block to which the skip mode OFF information is added is encoded using normal inter-screen prediction other than the skip mode. I do. The motion vector resulting from the encoding is stored in the motion vector memory 16 for reference in subsequent predictive encoding, and the decoded image signal of the encoded block is stored in the encoded image memory 15.

  FIG. 3 is a flowchart of the first embodiment.

  First, in step S101, an error ΔDC between the encoding target macroblock of the encoding target original image frame and the block of the immediately preceding original image frame is calculated and compared with a predetermined threshold Th1. As the error ΔDC, for example, the following absolute difference value of the sum of block luminance values is calculated.

ΔDC = | Σ _j Σ _i F _{m, n} ^(N) (i, j) −Σ _j Σ _i F _{m, n} ^(N−1) (i, j) |
Here, F _{m, n} ^(N) (i, j) is the (i, j) -th pixel value in the block at the (m, n) position of the Nth frame which is the original picture frame to be encoded, F _{m , n} ^(N−1) (i, j) is the luminance value of the (i, j) -th pixel in the block at the position (m, n) in the immediately preceding original image frame. Further, when the block size is 16 × 16 pixels, Σ _j represents the sum from j = 0 to j = 15, and Σ _i represents the sum from i = 0 to i = 15. Note that the error ΔDC is substantially equivalent even if an average error of the pixel values of the block is used.

  Next, in step S102, it is determined whether or not the error ΔDC calculated in step S101 is larger than a predetermined threshold value Th1, and if larger, the process proceeds to step S106. If it is smaller, the process proceeds to step S103.

  In step S103, a prediction error (SAD) between the encoding target macroblock and the block directly behind the previous encoded reference frame is calculated, and it is determined whether the prediction error is larger than a predetermined threshold Th2. If larger, the process proceeds to step S106, and if smaller, the process proceeds to step S104.

  In step S104, a predicted motion vector (PMV) is obtained from motion vectors of encoded blocks around the encoding target macroblock, and it is determined whether or not the predicted motion vector is zero. If it is not zero, the process proceeds to step S106, and if it is zero, the process proceeds to step S105.

  In step S105, the encoding target macroblock is set as a skip macroblock, and the skip mode is selected for encoding. On the other hand, in step S106, prediction encoding using motion search is performed without selecting a skip macroblock for the encoding target macroblock.

[Example 2]
The second embodiment is a method obtained by further improving the method of the first embodiment. FIG. 4 is a diagram for explaining the difference between the first embodiment and the second embodiment.

  When the prediction error between the encoding target frame and the encoded reference frame is actually calculated, the prediction error is large in the block with high variance even in the subjectively stationary region, and the prediction error is in the block with low variance. The error is reduced.

  In the first embodiment (method 1), as shown in FIG. 4A, a constant value is used as the threshold Th2 to be compared with the prediction error regardless of the magnitude of the variance. In the evaluation of the prediction error, a block having a prediction error smaller than the threshold Th2 is determined as a static region, and a block having a large error is determined as a moving region. , There may be a case where the stationary region is determined as a moving region in a highly dispersed portion.

  On the other hand, in the second embodiment (method 2), as shown in FIG. 4B, a large threshold Th2 is used when the variance is large, and a small threshold Th2 is used when the variance is small. The threshold value Th2 is switched depending on the size of the variance. As a result, it is possible to improve the accuracy of separation between the moving area and the stationary area, and to further reduce erroneous detection and missing of the skip mode target area.

  FIG. 5 shows a configuration example of the second embodiment of the skip mode control unit 1021. In FIG. 5, the configuration of the macroblock error calculation unit 20, the macroblock error evaluation unit 21, the prediction error evaluation unit 22, the prediction vector evaluation unit 23, the encoding unit 24, the motion vector memory 25, and the encoded image memory 26 is as follows. This is almost the same as that described in FIG.

  In the case of the second embodiment, a macroblock variance calculation unit 27 and a macroblock variance evaluation unit 28 are further provided. The macroblock variance calculation unit 27 calculates the variance (may be an activity) of pixel values of the encoding target macroblock, and notifies the macroblock variance evaluation unit 28 of the value. The macroblock variance evaluation unit 28 selects a large value as the threshold value Th2 when the variance is large, from among a plurality of threshold values having different sizes, and sets a small value when the variance is small. Select as the threshold Th2. This threshold value Th2 is transmitted to the prediction error evaluation unit 22. The prediction error evaluation unit 22 evaluates the prediction error using the threshold value Th2 set by the macroblock variance evaluation unit 28.

  FIG. 6 is a flowchart of the second embodiment.

  First, in step S201, an error ΔDC between the encoding target macroblock of the encoding target original image frame and the block of the immediately preceding original image frame is calculated and compared with a predetermined threshold Th1.

  Next, in step S202, it is determined whether or not the error ΔDC calculated in step S201 is larger than a predetermined threshold Th1, and if it is larger, the process proceeds to step S210. If it is smaller, the process proceeds to step S203.

  In step S203, a variance Var of luminance values of the encoding target macroblock is calculated. In step S204, it is determined whether or not the calculated variance Var is smaller than a predetermined threshold Th5.

  If the variance Var is smaller, the threshold value Th2 is set to a smaller predetermined value Th3 in step S205. If larger, the threshold Th2 is set to a larger predetermined value Th4 in step S206. Note that Th3 <Th4. Here, an example in which the threshold Th2 is selected from two types of values Th3 and Th4 having different sizes has been described. However, the threshold Th2 may be selected from three or more types according to the size of the variance Var. .

  In step S207, a prediction error (SAD) between the encoding target macroblock and the block immediately behind the previous encoded reference frame is calculated, and it is determined whether or not the prediction error is larger than a predetermined threshold Th2. If larger, the process proceeds to step S210, and if smaller, the process proceeds to step S208.

  In step S208, a predicted motion vector (PMV) is obtained from the motion vectors of the encoded blocks around the encoding target macroblock, and it is determined whether the predicted motion vector is zero. If it is not zero, the process proceeds to step S209, and if it is zero, the process proceeds to step S210.

  In step S209, the encoding target macroblock is set as a skip macroblock, and the skip mode is selected and encoded. On the other hand, in step S210, predictive encoding using motion search is performed without selecting a skip macroblock for the encoding target macroblock.

[Example 3]
The third embodiment is an example in which the encoding target frame is a B picture. The basic configuration is the same as the configuration example shown in FIG. FIG. 7 is a diagram for explaining the third embodiment, and FIG. 8 is a flowchart of the third embodiment.

  When the encoding target frame is a B-picture (Bi-predictive picture), the reference frame includes an encoded frame indicated by the L0 reference index and an encoded frame indicated by the L1 reference index as shown in FIG. Multiple. Accordingly, there are a plurality of original picture frames corresponding to the reference frame, such as an original picture frame corresponding to the L0 reference index and an original picture frame corresponding to the L1 reference index.

  Therefore, the total difference absolute value ΔDC of the block pixel values of the macroblock at the same position (true back) in the original picture frame corresponding to the encoded reference frame is also the same as that of the original picture frame corresponding to the L0 reference index. There are a plurality of ΔDC1 and ΔDC2 for the original image frame corresponding to the L1 reference index. In the third embodiment, for example, the larger one of ΔDC1 and ΔDC2 is used as the error compared with the threshold Th1. An average value may be used.

  The flow of processing will be described according to the flowchart shown in FIG.

  First, in step S301, the L0 reference index and the L1 reference index in the temporal / spatial direct mode are obtained, and the error ΔDC1 between the encoding target macroblock of the encoding target original image frame and the original block of the original image frame corresponding to the L0 reference index. (A difference absolute value of the sum of the block luminance values) is calculated, and an error ΔDC2 between the encoding target macroblock of the encoding target original image frame and the true back block of the original image frame corresponding to the L1 reference index is calculated.

  Next, in step S302, it is determined whether the larger error of the errors ΔDC1 and ΔDC2 calculated in step S301 is larger than a predetermined threshold value Th1, and if larger, the process proceeds to step S306. If it is smaller, the process proceeds to step S303.

  In step S303, the prediction error SAD1 between the encoding target macroblock and the true back block of the encoded reference frame of the L0 reference index is calculated, and the prediction error SAD2 of the true reference block of the encoded reference frame of the L1 reference index is calculated. Calculate The larger one of the prediction errors SAD1 and SAD2 is compared with a predetermined threshold Th2. If the prediction error is larger, the process proceeds to step S306, and if smaller, the process proceeds to step S304.

  In step S304, it is determined whether the direct vector is zero. If it is not zero, the process proceeds to step S306, and if it is zero, the process proceeds to step S305.

  In step S305, the encoding target macroblock is set as a skip macroblock, and the skip mode is selected for encoding. On the other hand, in step S306, prediction encoding using motion search is performed without selecting a skip macroblock for the encoding target macroblock.

[Example 4]
In the fourth embodiment, in the encoding in the case of a B picture (Bi-predictive Picture) similar to the third embodiment, the threshold value Th2 used in the determination in step S303 is set as the pixel value of the encoding target block, as in the second embodiment. In this example, a large value is set when the variance is large, and a small value is set when the variance is small.

  The flowchart of Example 4 is shown in FIG. In FIG. 9, the processes of steps S401 to S402 and steps S407 to S410 are the same as the processes of steps S301 to S302 and steps S303 to S306 described in FIG. Further, the processing in steps S403 to S406 in FIG. 9 is the same as the processing in steps S203 to S206 described in FIG.

[Example 5]
The fifth embodiment is an example in which the image to be encoded is an interlaced image and is field encoded (P picture). FIG. 10 is a diagram for explaining the fifth embodiment. An image consists of a top field (Top field) and a bottom field (Bottom field), and each field is encoded. As shown in FIG. 10, the error ΔDC and the prediction error SAD with respect to the macroblock at the same position in the field of the same parity of the encoding target macroblock are as follows. Used.

  First, in step 1, an error ΔDC (for example, the absolute difference value of the sum of block luminances) between the encoding target macroblock of the encoding target original field and the block at the same position of the previous original field of the same parity is calculated, To the threshold value Th1.

  In the procedure 2, if the error ΔDC calculated in the procedure 1 is larger than the threshold Th1, the motion search is performed without selecting the skip mode. If it is smaller than the threshold value Th1, the prediction error SAD between the encoding target macroblock and the true back block of the encoded reference field immediately before the same parity is calculated. If the prediction error SAD is smaller than the threshold Th2 and the predicted motion vector (PMV) is zero, the skip mode is selected, and if it is greater than the threshold Th2 or the predicted motion vector (PMV) is not zero, the skip is performed. Perform motion search without selecting a mode.

  In the fifth embodiment, the threshold Th2 may be switched according to the distribution of the pixel values of the encoding target macroblock, as in the second and fourth embodiments.

  The moving image encoding process described above can be realized by a computer and a software program, and the program can be recorded on a computer-readable recording medium or provided through a network.

  FIG. 11 shows an example of a hardware configuration when the moving image encoding apparatus is configured by a computer and a software program. This system includes a CPU 50 that executes a program, a memory 51 such as a RAM that stores programs and data accessed by the CPU 50, and a video signal input unit 52 that inputs a video signal to be encoded from a camera or the like (disk device). Or a storage unit that stores a video signal such as a video program 531 that is a software program that causes the CPU 50 to execute the encoding process described in the embodiment of the present invention; The encoded stream generated by the CPU 50 executing the moving image encoding program 531 loaded in the memory 51, for example, is output via a network. The encoded stream output unit 54 (stores the encoded stream by the disk device or the like). Connected by a bus. Going on.

  As mentioned above, although embodiment of this invention was described with reference to drawings, the said embodiment is only the illustration of this invention, and it is clear that this invention is not limited to the said embodiment. Accordingly, additions, omissions, substitutions, and other modifications of the components may be made without departing from the spirit and technical scope of the present invention.

1021 Skip mode control unit 10 Macroblock error calculation unit 11 Macroblock error evaluation unit 12 Prediction error evaluation unit 13 Prediction vector evaluation unit 14 Coding unit 15 Encoded image memory 16 Motion vector memory

Claims (7)

In a video encoding method having a skip mode as a prediction mode in predictive encoding,
When the encoding target macroblock is a bi-predictive image, the total of the block pixel values of the encoding target macroblock and the encoding in the original picture frame corresponding to the first encoded reference frame in the temporal / spatial direct mode The error of the block pixel value of the macroblock at the same position as the target macroblock or the block pixel value of the macroblock at the same position as the target macroblock in the original picture frame corresponding to the second encoded reference frame The process of calculating the larger error as the first error among the errors with the total ,
Comparing a predetermined first threshold value with the first error, and setting the skip mode off if the first error is greater;
When the first error is less than or equal to the first threshold, the error between the encoding target macroblock and the reference block in the first encoded reference frame in the temporal / spatial direct mode or the second encoded reference A process of calculating a larger error as a second error among errors from the reference block in the frame ;
Comparing a predetermined second threshold value with the second error, and setting the skip mode off if the second error is greater;
Whether the motion vector or predicted motion vector of the encoding target macroblock is zero or less than a predetermined value when the first error is less than the first threshold and the second error is less than the second threshold A process of setting the skip mode on if it is zero or less than the predetermined value, and setting the skip mode off if not,
A moving picture encoding method comprising: encoding with a skip mode when the skip mode is on, and encoding with a prediction mode other than the skip mode when the skip mode is off. In a video encoding method having a skip mode as a prediction mode in predictive encoding,
When the encoding target macroblock is a bi-predictive image, the total of the block pixel values of the encoding target macroblock and the encoding in the original picture frame corresponding to the first encoded reference frame in the temporal / spatial direct mode The error of the block pixel value of the macroblock at the same position as the target macroblock or the block pixel value of the macroblock at the same position as the target macroblock in the original picture frame corresponding to the second encoded reference frame The process of calculating the larger error as the first error among the errors with the total ,
Comparing a predetermined first threshold value with the first error, and setting the skip mode off if the first error is greater;
When the first error is less than or equal to the first threshold, the error between the encoding target macroblock and the reference block in the first encoded reference frame in the temporal / spatial direct mode or the second encoded reference A process of calculating a larger error as a second error among errors from the reference block in the frame ;
A process of calculating a variance of pixel values of a macroblock to be encoded;
Selecting and setting a second threshold value according to the variance value from a plurality of threshold values of different sizes;
Comparing the second threshold with the second error, and setting the skip mode off if the second error is greater;
Whether the motion vector or predicted motion vector of the encoding target macroblock is zero or less than a predetermined value when the first error is less than the first threshold and the second error is less than the second threshold A process of setting the skip mode on if it is zero or less than the predetermined value, and setting the skip mode off if not ,
A moving picture encoding method comprising: encoding with a skip mode when the skip mode is on, and encoding with a prediction mode other than the skip mode when the skip mode is off. In the moving image encoding method according to claim 1 or 2 ,
For field encoding,
In the process of calculating as the first error , instead of the process of calculating the first error, the total block pixel value of the encoding target macroblock and the encoding target macroblock in the temporal / spatial direct mode are the same. An error in the original picture field corresponding to the first encoded reference field of parity with the sum of the block pixel values of the macroblock located at the same position as the encoding target macroblock or the original picture corresponding to the second encoded reference field Calculating an error from the sum of the block pixel values of the macroblock at the same position as the encoding target macroblock in the field ;
In the process of calculating the second error, instead of the process of calculating the second error, the error between the encoding target macroblock and the reference block in the first encoded reference field in the temporal / spatial direct mode. Alternatively, the moving picture coding method characterized by calculating an error from the reference block in the second coded reference field . In a video encoding apparatus having a skip mode as a prediction mode in predictive encoding,
When the encoding target macroblock is a bi-predictive image, the total of the block pixel values of the encoding target macroblock and the encoding in the original picture frame corresponding to the first encoded reference frame in the temporal / spatial direct mode The error of the block pixel value of the macroblock at the same position as the target macroblock or the block pixel value of the macroblock at the same position as the target macroblock in the original picture frame corresponding to the second encoded reference frame A first error calculating means for calculating a larger error as a first error among errors from the total ,
A first evaluation means for comparing a predetermined first threshold value with the first error, and setting the skip mode off when the first error is larger;
When the first error is less than or equal to the first threshold, the error between the encoding target macroblock and the reference block in the first encoded reference frame in the temporal / spatial direct mode or the second encoded reference A second error calculating means for calculating a larger error as a second error among errors from the reference block in the frame ;
A second evaluation means for comparing a predetermined second threshold value with the second error and setting the skip mode to off when the second error is larger;
Whether the motion vector or predicted motion vector of the encoding target macroblock is zero or less than a predetermined value when the first error is less than the first threshold and the second error is less than the second threshold A third evaluation means for setting the skip mode on if it is zero or less than the predetermined value, and setting the skip mode off if not;
Coding means for performing coding in skip mode when skip mode is on, and coding in prediction modes other than skip mode when skip mode is off apparatus. In a video encoding apparatus having a skip mode as a prediction mode in predictive encoding,
When the encoding target macroblock is a bi-predictive image, the total of the block pixel values of the encoding target macroblock and the encoding in the original picture frame corresponding to the first encoded reference frame in the temporal / spatial direct mode The error of the block pixel value of the macroblock at the same position as the target macroblock or the block pixel value of the macroblock at the same position as the target macroblock in the original picture frame corresponding to the second encoded reference frame A first error calculating means for calculating a larger error as a first error among errors from the total ,
A first evaluation means for comparing a predetermined first threshold value with the first error, and setting the skip mode off when the first error is larger;
When the first error is less than or equal to the first threshold, the error between the encoding target macroblock and the reference block in the first encoded reference frame in the temporal / spatial direct mode or the second encoded reference A second error calculating means for calculating a larger error as a second error among errors from the reference block in the frame ;
A variance calculation means for calculating the variance of pixel values of the encoding target macroblock;
A variance evaluation means for selecting and setting a second threshold value according to the variance value from a plurality of threshold values of different sizes;
A second evaluation means for comparing the second threshold value with the second error and setting the skip mode to off when the second error is larger;
Whether the motion vector or predicted motion vector of the encoding target macroblock is zero or less than a predetermined value when the first error is less than the first threshold and the second error is less than the second threshold A third evaluation means for setting the skip mode on if it is zero or less than the predetermined value, and setting the skip mode off if not;
Coding means for performing coding in skip mode when skip mode is on, and coding in prediction modes other than skip mode when skip mode is off apparatus. In the moving image encoding device according to claim 4 or 5 ,
For field encoding,
The first error calculation means replaces the process of calculating the first error with the sum of block pixel values of the encoding target macroblock and the same parity as the encoding target macroblock in the temporal / spatial direct mode. In the original image field corresponding to the second encoded reference field, or an error from the sum of the block pixel values of the macroblock located at the same position as the encoding target macroblock in the original image field corresponding to the first encoded reference field Calculating an error from the sum of block pixel values of a macroblock at the same position as the macroblock to be encoded ;
The second error calculation means replaces the process of calculating the second error with an error or first error between the encoding target macroblock and the reference block in the first encoded reference field in the temporal / spatial direct mode. An apparatus for encoding a moving picture, comprising: calculating an error from a reference block in two encoded reference fields . A moving picture coding program for causing a computer to execute the moving picture coding method according to any one of claims 1 to 3 .

Images