JPH0818952A - Image signal coder - Google Patents

Abstract

PURPOSE:To reduce the information quantity and to improve the image quality by setting properly quantization steps when quantization control is conducted adaptively depending on local property of an image in the high efficiency coding of a digital image signal. CONSTITUTION:A feature amt. is calculated for each small divided block. A frame feature amt. distribution state calculation means 3 checks the distribution state of a feature amt. of frames, and a frame feature amt. distribution state storage means 4 stores a frame feature amt. distribution state of plural preceding and succeeding frames. A current frame feature amt. distribution state estimate means 5 estimates a frame feature amt. distribution state of a coded object frame based on the frame feature amt. distribution states of the plural preceding or succeeding frames to be stored, and a quantization characteristic decision means 6 refers to the estimate to decide a quantization characteristic of coded object small blocks.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention adaptively allocates a limited amount of information according to the local property of an image in high-efficiency coding of a television signal, etc. The present invention relates to an image signal encoding device for improving.

[0002]

2. Description of the Related Art In high-efficiency coding of image signals such as television signals, areas in which coding noise is easily detected are quantized in fine quantization steps, and areas in which coding noise are difficult to detect are coarse quantization steps. The method of quantizing is often used.

FIG. 5 is a diagram for explaining an example of a conventional encoding device. In this example, it is assumed that standard motion compensation and discrete cosine transform are used as the image encoding device. Also, as one of the human visual characteristics, a flat area where the pixel value changes little is noticeable when the pixel value fluctuates even a little, and a region where the pixel value changes drastically is inconspicuous even if the pixel value fluctuates slightly. Because it is effective,
Consider a case where the variance of pixel values in a frame is used as a feature amount for determining a quantization step.

First, the image input from the input terminal 901 is divided into N × N small blocks by the small block dividing unit 902. A motion vector 905 is obtained in the motion vector detection unit 904 from the image signal 903 of the small block and the image signal of the previous frame stored in the frame memory 914. Based on the motion vector 905, the motion compensation unit 906 Motion compensation is performed and subtractor 9
After the motion compensation prediction error signal 908 is obtained at 07,
The discrete cosine transform unit 909 performs discrete cosine transform on the motion compensation prediction error signal 908.

On the other hand, an image signal 90 divided into small blocks
For 3, the intra-frame variance calculation unit 915 obtains the intra-frame variance 916 by the following calculation and sends it to the quantization step calculation unit 917.

V = Σ _j Σ _i (s _ij −s _M ) ² / (N × N) where s _ij is the pixel value in the small block, and s _M is the average thereof. Further, Σ _j represents the total sum from j = 1 to N, and Σ _i represents the total sum from i = 1 to N.

Next, in the quantization step calculation unit 917, instead of the average value of the intra-frame variance of the encoding target frame, the average value of the intra-frame variance of the frame encoded immediately before the encoding target frame _V'M Is used to determine the magnitude of the intra-frame variance (V) 916 obtained for each small block. The parameter A representing this magnitude relation is
Calculated from the following formula.

A = (2 · V + V ′ _M ) / (V + 2 · V ′ _M ) A is the value of the frame encoded by the intra-frame variance (V) 916 obtained for each small block immediately before the encoding target frame. When the average value of the intra-frame variance is smaller than V ′ _M , the minimum value is 0.5 and the value is smaller than 1.0.
When 916 is larger than the average value V ′ _M of the intra-frame variance of the frame encoded immediately before the encoding target frame, the maximum value is 2.0 and the value is larger than 1.0.

Then, the quantization step Q obtained from the buffer memory occupation amount 918 for outputting the encoded data at a constant bit rate is corrected by A, and the quantization step Q _act actually used is determined. .

Q _act = A · Q By correcting the quantization step Q obtained from the buffer memory occupation amount 918 with A, a coarse quantum is obtained when the intra-frame variance (V) 916 obtained for each small block is large. The process of setting the conversion step and finely setting the small part is performed.

The determined quantization step (Q _act ) 91
9, the discrete cosine transform coefficient 920 is used by the quantizer 9
After being quantized by 10, variable-length coding is performed by the variable-length coding unit 921, it is input to the buffer memory 922 and is output to the encoded data output terminal 923 at a constant bit rate.

The quantized discrete cosine transform coefficient is inversely quantized by an inverse quantizer 911 and inverse discrete cosine transform by an inverse discrete cosine transform unit 912, and then added with the motion-compensated previous frame data. It is added in the unit 913 and stored in the frame memory 914 for motion compensation prediction of the next frame. Further, the intra-frame variance average calculation unit 92
In 4, the average value of the intra-frame variance of the frame to be encoded is calculated and used as an estimated value of the average value of the intra-frame variance of the frame to be encoded next.

According to such a method, a portion having a flat pixel value can be finely quantized and a portion having a large change can be roughly quantized, so that a limited amount of information can be distributed according to the masking effect. As a result, the quality of the encoded image can be improved. In addition, since the average value of the intra-frame variance of the frame encoded immediately before is used as the predicted value of the average value of the intra-frame variance of the encoding target frame, the intra-frame variance of the encoding target frame is calculated before encoding. It is not necessary to obtain the average value of

[0014]

In the conventional method, the distribution state of the feature amount of the immediately preceding frame is used as the estimated value as the distribution state of the feature amount of the encoding target frame. When the distribution state of the feature quantity significantly changes in, there is a problem that the estimation is missed because the distribution state of the feature quantity is greatly different between the frame immediately before encoding and the encoding target frame.

As an example, FIG. 6 shows a case where the average value of the intra-frame variance greatly changes between frames. As shown in FIG. 6, the quantization control in the case where the distribution state of the feature amount greatly changes between frames is performed as follows.

First, as shown in FIG. 7A, when the average value of the intra-frame variance of the immediately preceding coded frame used as an estimated value is smaller than the average value of the intra-frame variance of the frame to be coded, If the average value of the intra-frame variance of the frame to be coded is used, a part of the small blocks that should be set so that fine quantization is performed is an estimated value of the intra-frame variance of the immediately preceding frame. Coarse quantization may be performed by using the average value.

On the contrary, as shown in FIG. 7B, the average value of the intra-frame variance of the immediately preceding coded frame used as an estimated value is larger than the average value of the intra-frame variance of the frame to be coded. In this case, if the average value of the intra-frame variance of the frame to be encoded is used, a part of the small blocks that should be set so that coarse quantization is performed is an estimated value within the frame of the immediately preceding frame. Fine quantization may be performed by using the average value of the variances.

As described above, in the case of an image signal in which the distribution state of the feature amount of the immediately preceding encoded frame and the distribution state of the feature amount of the encoding target frame are significantly different, an inappropriate quantization step is set, There is a problem that the quantization control does not work properly.

[0019]

The present invention has been made to solve this problem, and is provided with the means described below.

FIG. 1 is a block diagram of the principle of the present invention.
First, the small block dividing means 1 inputs a digital image signal such as a television signal and divides the image data of the encoding target frame into small blocks. The small block feature amount calculation means 2 calculates a feature amount representing the property of each of the divided small blocks. The frame feature amount distribution state calculation means 3 checks the distribution state of the feature amount over one frame from the result of the small block feature amount calculation means 2, and stores the result in the frame feature amount distribution state storage means 4.
The frame feature amount distribution state storage means 4 stores the feature amount distribution states of a plurality of past or future frames.

The current frame feature quantity distribution state estimating means 5 is
The distribution state of the feature quantity of the encoding target frame is estimated from the distribution states of the feature quantities of a plurality of past or future frames stored in the frame feature quantity distribution state storage unit 4. The quantization characteristic determining unit 6 refers to the estimated distribution of the feature amount of the encoding target frame and the feature amount obtained by the small block feature amount calculating unit 2 for each small block in the encoding target frame. Thus, the quantization characteristic of the small block to be encoded is determined.

The quantizing means 7 quantizes each small block according to the quantizing characteristic determined by the quantizing characteristic determining means 6. The variable length coding means 8 codes the result.

[0023]

In the present invention, the distribution state of the characteristic quantities of a plurality of past or future frames encoded before the encoding target frame is stored. Next, as shown in FIG. 2, the estimated distribution value of the characteristic amount of the encoding target frame is estimated using the stored characteristic amount distribution states of a plurality of past or future frames.

D' _n = f (D _ni , ..., D _n-1 ,
D _{n + 1} , ..., D _{n + j} ) Here, D ′ _n is a distribution state estimation value of the feature amount of the encoding target frame, and D _k is a distribution of the feature amount of the already encoded past or future frame. The states, i, and j respectively represent the past and future number of frames used for estimating the feature amount distribution state of the encoding target frame.

With this configuration, when estimating the distribution of the feature quantity of the coding target frame, the error between the distribution of the feature quantity of the coding target frame and the estimated value can be reduced.

[0026]

FIG. 3 is a diagram showing a block configuration of an embodiment of the present invention. In this embodiment, MPE is used as an image encoding method.
Using motion compensation + discrete cosine transform coding as used in G2 etc., using the intra-frame variance as the feature quantity and the average value of the entire frame as the index of the distribution state, the average value of the luminance variance of the coding target frame It is assumed that is estimated by linear approximation from the average value of the intraframe variances of the two frames encoded immediately before. Further, it is assumed that a frame using only motion compensation from the past (hereinafter referred to as P picture) and a frame using motion compensation from past and future (hereinafter referred to as B picture) are encoded with the following structure. .

B _n-2 P _n-1 B _n-1 P _n B _n P _{n + 1} B _{n + 1} P _{n + 2} B _{n + 2} P _{n + 3} ... Also, the actual encoding order is It is as follows. ... P _n-1 B _n-2 P _n B _n-1 P _{n + 1} B _n P _{n + 2} B _{n + 1} P _{n + 3} B _{n + 2} First, the image input from the input terminal 101 is The small block dividing unit 102 divides into N × N small blocks.
A motion vector 105 is obtained by the motion vector detection unit 104 from the image signal 103 of a small block and the image signal of one frame before stored in the frame memory 114. Based on this motion vector 105, the motion compensation unit 1
After motion compensation is performed at 06, the motion compensation prediction error signal 108 is obtained by the subtractor 107, and then the discrete cosine transform unit 109 performs discrete cosine transform on the motion compensation prediction error signal 108.

On the other hand, the image signal 10 divided into small blocks
With respect to 3, the intraframe variance calculation unit 115 obtains the intraframe variance 116 by the following calculation and sends it to the quantization step calculation unit 117.

V = Σ _j Σ _i (s _ij −s _M ) ² / (N × N) where s _ij is the pixel value in the small block, and s _M is the average thereof. Further, Σ _j represents the total sum from j = 1 to N, and Σ _i represents the total sum from i = 1 to N.

Subsequently, the intra-frame variance average value estimation unit 12
5, in-frame variance average value memories 127 to 12
Average values D _n−1 , D of intra-frame variances of the two frames encoded immediately before the encoding target frame stored in
_{From n-2} , the average value of the intra-frame variance of the coding target frame is estimated by linear approximation. This linear approximation is performed as follows.

First, in the case of a P picture, _assuming that the encoding target frame is P _n , the immediately preceding two frames are D _n-2 = P _n-1 and D _n-1 = B _n-2. Becomes Therefore, D _Pn is linearly approximated and estimated from the average value of the intra _- frame variances of P _n-1 and B _n-2 . FIG. 4A shows a method of estimating the average value of the intraframe variance in the case of P pictures. As is clear from the figure, the estimated value D' _Pn is obtained from the following equation.

D' _Pn = _{-2.D Bn-2} + _{3.D Pn-1 In} the case of a B picture, the encoding target frame is B
_{If n} , the immediately preceding two frames are D _n-2 =
B _n-1 and D _n-1 = P _{n + 1} . Therefore, this B
D _Bn is linearly approximated and estimated from the average value of the intra-frame variances of _n−1 and P _{n + 1} . FIG. 4B shows a method of estimating the average value of the intraframe variance in the case of a B picture. As is clear from the figure, the estimated value D' _Bn is obtained from the following equation.

D '_Bn= (D_Bn-1+ 2 · D_{Pn + 1}) / 3 Next, in the quantization step calculation unit 117, the frame
The encoding target frame estimated by the inner variance average value estimation unit 125
Estimated value (D ' _n) 12
Using 6 to find the inside of the frame for each small block
The magnitude of the dispersion V is determined. Parameters that express this magnitude relationship
Data A is calculated by the following equation.

A = (2 · V + D ′ _n ) / (V + 2 · D ′ _n ) A is the estimation of the intra-frame variance average value of the target frame for which the intra-frame variance (V) 116 obtained for each small block is If the value is smaller than (D ' _n ) 126,
The minimum value of 0.5 is smaller than 1.0, and the intra-frame variance (V) 116 obtained for each small block is calculated from the estimated intra-frame variance average value (D ′ _n ) 126 of the encoding target frame. If it is larger, the maximum value is 2.0 and the value is larger than 1.0. The quantization step Q obtained from the buffer memory occupation amount 118 is corrected by this A, and the quantization step (Q _act ) actually used
119 is determined.

Q _act = A · Q By correcting the quantization step Q obtained from the buffer memory occupancy amount 118 with A, a portion having a large intra-frame variance (V) 116 obtained for each small block is a rough quantum. The process of setting the conversion step and finely setting the small part is performed.

The determined quantization step (Q _act ) 11
9, the discrete cosine transform coefficient 120 is used by the quantizer 1
After being quantized by 10 and variable-length coded by the variable-length coding unit 121, it is input to the buffer memory 122 and output to the encoded data output terminal 123 at a constant bit rate.

The quantized discrete cosine transform coefficient is inversely quantized by the inverse quantizer 111, inverse discrete cosine transform by the inverse discrete cosine transform unit 112, and then added with the motion-compensated previous frame data. It is added by the device 113 and stored in the frame memory 114 for motion compensation prediction of the next frame.

Further, when the frame head detecting section 129 detects the input of the frame to be encoded, the switches 130 to 131 are closed, and the encoding is performed two frames before stored in the intra-frame variance average value memory 127. The intra-frame variance average of the thus-generated frames is transferred to the intra-frame variance average value memory 128, and the intra-frame variance average value memory 127 stores the immediately preceding encoded frame calculated by the intra-frame variance average value calculation unit 124. The intraframe variance average value is stored.

In the embodiment described above, the image signal coding method is motion compensation + discrete cosine transform coding, the feature quantity is the intra-frame variance, and the index of the distribution state of the feature quantity is the average value. is not. Further, the past or future frames used for estimation and the number thereof, the method of estimating the distribution state of the feature amount of the encoding target frame, and the method of calculating the quantization control parameter can be arbitrarily determined in the implementation of the present invention.

[0040]

As described above, according to the present invention,
When adaptively performing quantization control on the local features of the image signal, it is possible to divide into regions where deterioration is prominent and regions where deterioration is not conspicuous without obtaining the distribution state of the feature quantities of the encoding target frame in advance. On the other hand, it becomes possible to set an appropriate quantization step. As a result, in a region where deterioration is apt to be noticed, deterioration can be made inconspicuous by performing fine quantization, and conversely, in a region in which deterioration is not noticeable, the amount of information can be reduced by performing coarse quantization, so that a certain amount of information can be obtained. Visual deterioration can be reduced based on the amount of information, and image quality can be improved.

[Brief description of drawings]

FIG. 1 is a principle block diagram of the present invention.

FIG. 2 is a diagram illustrating an estimated value of a distribution state of feature quantities of a frame to be encoded.

FIG. 3 is a block diagram of an embodiment of the present invention.

FIG. 4 is a diagram illustrating a method of estimating an average value of intra-frame variance.

[Fig. 5] Fig. 5 is a diagram illustrating a configuration example of a conventional encoding device.

FIG. 6 is a diagram showing changes in distribution of feature quantities and estimated values.

FIG. 7 is a diagram showing an influence of a prediction error of a distribution state of feature quantities on quantization control.

[Explanation of symbols]

 1 small block dividing means 2 small block feature amount calculating means 3 frame feature amount distribution state calculating means 4 frame feature amount distribution state storing means 5 current frame feature amount distribution state estimating means 6 quantization characteristic determining means 7 quantizing means 8 variable length Encoding means 101 Input terminal 102 Small block division unit 103 Small block image signal 104 Motion vector detection unit 105 Motion vector 106 Motion compensation unit 107 Subtractor 108 Motion compensation prediction error signal 109 Discrete cosine transform unit 110 Quantization unit 111 Inverse quantum Optimizer 112 Inverse Discrete Cosine Transform 113 Adder 114 Frame Memory 115 Intraframe Variance Calculator 116 Intraframe Variance 117 Quantization Step Calculator 118 Buffer Memory Occupancy 119 Quantization Step 120 Discrete Cosine Transform Coefficient 121 Variable Length Coding 1 22 buffer memory 123 output terminal 124 intra-frame variance average value calculation unit 125 intra-frame variance average value estimation unit 126 intra-frame variance average value estimation value 127 to 128 intra-frame variance average value memory 129 frame head detection unit 130 to 131 switch

Claims (1)

[Claims] 1. An image signal encoding apparatus for encoding a digital image signal by changing a quantization characteristic according to a distribution state of feature amounts of image data, and dividing the image data of an encoding target frame into small blocks. Means, a means for obtaining a feature amount representing the property of the small block for each small block, a means for checking and storing the distribution state of the feature amount over one frame, and a feature of a plurality of past or future frames Means for estimating the distribution state of the feature amount of the encoding target frame from the distribution state of the amount, and the estimated distribution state of the feature amount of the encoding target frame and the feature amount obtained for each small block in the encoding target frame By referring to
An image signal coding apparatus, comprising: means for determining a quantization characteristic of a small block to be coded.