JPH0750837A - Motion picture coder - Google Patents

Abstract

PURPOSE:To suppress the production of a code quantity by not coding a high frequency component of an orthogonal transformation coefficient depending on a mean value and a variance of a prediction error. CONSTITUTION:A picture signal of a current frame and a reproduced picture of a preceding frame are compared and a motion vector detection section 2 calculates a motion vector and a motion compensation prediction control signal and a motion compensation prediction section 4 applies motion compensation prediction to a reproduced picture of the preceding frame. A coding method discrimination section 5 discriminates whether a coded block is subject to inter-frame coding or an in-frame coding. A prediction error calculation section 7 takes a difference between an original picture signal of a coded block and a prediction signal to calculate a prediction error. An orthogonal transformation section 9 implements orthogonal transformation and a coefficient and replacement section 10 rearranges the orthogonal transformation coefficients in the order of frequencies. A mean value and variance calculation section 12 calculates the mean value and the variance of the total sum of difference absolute values calculated by the prediction error calculation section 7. A coding coefficient selection section 13 selects a low frequency component of the orthogonal transformation coefficient coded by the mean value and variance and a high frequency component not coded.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture coding apparatus for coding a moving picture signal by digital processing, and more particularly to a motion compensation predictive interframe coding apparatus.

[0002]

2. Description of the Related Art Conventionally, in an orthogonal transform coding apparatus for orthogonally transforming an original signal and quantizing its transform coefficient for transmission,
A region that is flatter than the variance value is determined, and coarse quantization is performed on the block by increasing the quantization step width, or a process is performed in which a threshold value is set and coding is not performed when the threshold value is less than the threshold value. Was there.

[0003]

In the prior art, when coarse quantization is performed, an error due to quantization may occur even in a low frequency component containing important information, and block distortion may occur. There was a problem. An object of the present invention is to find a flat region of an image in a wide range from the average and variance of the prediction error signal, and cut the high frequency component of the orthogonal transform coefficient rearranged in frequency order in the block of the region. An object of the present invention is to provide a moving picture signal coding device capable of suppressing the generation of an extra code amount and obtaining an efficient compression ratio by increasing the number of runs of zeros of quantized data.

[0004]

DISCLOSURE OF THE INVENTION The present invention is to achieve the above-mentioned object, and after each frame of an input moving image signal is digitized, an orthogonal transformation is performed in each block for dividing a screen. A moving picture coding apparatus that quantizes, quantizes the quantized digital data, and performs variable length coding to transmit the quantized digital data via a transmission buffer. Prediction means for predicting the current image to obtain a prediction value, means for subtracting the prediction value from the current image to generate a prediction error signal, and inter-frame coding or intra-frame coding for each block A coding method determining means for determining whether or not, a motion vector detecting means for calculating a motion vector, a motion compensating means for motion compensation using the motion vector, and a motion vector calculating means for obtaining the motion vector. Average variance calculating means for calculating the average and variance of the sums within the search range of the pair difference values, coefficient rearranging means for rearranging the coefficients in frequency order to increase the number of runs of zeros with respect to the orthogonal transform coefficient, and the average And a means for replacing the coefficient rearranged by variance with 0.

[0005]

When the average of the sums of the absolute values of the differences with the block within the search range of the previous frame obtained at the time of motion vector detection is close to zero and the variance is small, the difference with the surroundings of the same position of the previous frame is Because the image is small, it can be said that the image in that range is a flat area. The fact that the error occurs even in the flat area is often due to the noise generated especially in the video system. If both the average and the variance are zero, there is no difference within the search range, that is, there is no difference within the search range, so it can be said that the images in that range are flat and identical. Therefore, it can be said that the average is close to zero, and the smaller the variance, the closer to a flat image. When determining the number to be encoded in the block from the average and variance, for orthogonal transform coefficients rearranged in frequency order, the closer the average and variance are to zero, reduce the number of low-frequency components to be encoded, A large range of high frequency components to be replaced with zero is taken. On the contrary, when the average and the variance are large, it can be said that flat regions and edge regions are mixed. Therefore, the high frequency component becomes important and it is necessary to encode all the coefficients. From the above, the flat region is determined by the average and variance of the sum of absolute differences, and the orthogonal transform coefficient that does not encode the high frequency component in the block is determined by the degree of the average and variance determined as the flat region. By suppressing the noise of the high frequency component, the generated code amount is suppressed, and as a result, smooth motion can be reproduced.

[0006]

Embodiments of the moving picture coding apparatus according to the present invention will be described below. In FIG. 1, 1 is an input terminal, 2 is a motion vector detection unit that compares a current frame image signal with a reproduced image of a previous frame to calculate a motion vector and a motion compensation prediction control signal, and 3 is a current frame and a previous frame. Frame memory unit for accumulating the reproduced image of 4 is a motion compensation prediction unit for performing motion compensation prediction for the reproduced image of the previous frame, and 5 is a code for determining whether the block to be encoded is inter-frame coded or intra-frame coded. An encoding method determination unit, 6 is an in-loop filter unit that performs two-dimensional low-pass filter processing on a motion compensation prediction signal, and 7 is a prediction error calculation unit that calculates the difference between the original image signal of the coding block and the prediction signal and calculates the prediction error. , 8
Is a switch unit for selecting a signal to be orthogonally transformed by a coding system selection signal and a signal for calculating a reproduced image, 9 is an orthogonal transformation unit for performing orthogonal transformation, and 10 is a coefficient for rearranging orthogonal transformation coefficients in frequency order. FIG. 3 is a rearrangement unit.
Shows the order of rearrangement when the block is 4 × 4. 1
1 is a quantizer that quantizes the orthogonal transform coefficients rearranged in frequency order; 12 is an average variance calculator that calculates the average and variance of the sums of the absolute differences calculated by the prediction error calculator 7;
13 is a coding coefficient selection unit that selects a low-frequency component of the orthogonal transform coefficient to be coded based on the average and variance and a high-frequency component that is not coded, and 14 is a coefficient that the coding coefficient selection unit 13 has determined not to code. This is a high-frequency component removing unit for replacing with zero, and the encoding coefficient selecting unit 13 shown in FIG.
A state in which high frequency components are removed by selecting individual pieces is shown. Reference numeral 15 is a variable length coding unit for variable length coding the quantized orthogonal transform coefficient, 16 is a transmission buffer memory unit for temporarily storing transmission data, 17 is a dequantization unit for dequantizing a quantized signal, 18 Is an inverse reordering unit that restores the original order by reversing the reordering performed in Step 10, 19 is an inverse orthogonal transform unit that performs an inverse orthogonal transform of the dequantized signal, and 20 is a playback that calculates a playback image of the current frame. The image calculation unit 21 is an output terminal for outputting encoded data.

The operation of the above arrangement will be described below. The signal input from the input terminal 1 is compared with the reproduced image of the previous frame read from the frame memory unit 3 by the motion vector detection unit 2 to calculate a motion vector. Further, the motion vector detection unit 2 determines whether the motion compensation prediction for the encoded block is valid or invalid, and outputs the result to the motion compensation prediction unit 4 as motion compensation prediction control information.
The motion-compensated prediction unit 4 predicts a motion vector when performing motion-compensated prediction with a motion vector signal on a reproduced image of a previous frame at the same position as the coding block, and when motion-compensated prediction is not performed, the motion-compensated prediction is performed without doing anything. Output as a signal. The motion vector is a block (N × N
The value of the upper left corner (NK, NL) of the pixel) is set to Xm (NK, NL)
), The sum of the absolute values of the differences between the block in the previous frame and the block in the current frame whose positions are shifted by i, j,

[0008]

[Equation 1]

The motion vector is (i, j) that minimizes.

The S output from the motion vector detecting section
Calculate the mean and variance of ij. The coding method determination unit 5 compares the input television signal and the motion-compensated prediction signal on a block-by-block basis to determine the effectiveness of the motion-compensated prediction. If the effectiveness of the motion-compensated prediction is small, the intra-frame code is valid for the block. If the validity is high, the interframe code is determined to be valid and the result is output as a coding method selection signal. The in-loop filter unit 6 performs in-loop filter processing, which is two-dimensional low-pass filter processing, on the motion-compensated prediction signal with respect to the motion-compensated signal and when the coding block performs motion-compensated prediction, and outputs the predicted signal. The prediction error calculation unit 7 takes the difference between the input signal of the coding block and the prediction signal, and outputs the result as a prediction error signal. The switch unit 8 selects the input television signal as a signal to be orthogonally converted when the coded signal selection signal is intra, and selects the prediction error signal when inter is selected. In most cases, the orthogonal transform unit 9 uses discrete cosine transform.

The average variance calculator 12 calculates the average value μ and the variance value σ ² as

[0012]

[Equation 2]

Then, the coding coefficient selection unit 13 creates a coding selection signal for selecting a coefficient to be coded in the block and a coefficient not to be coded, based on the average value and the variance value. The orthogonal transform coefficient rearranged in the frequency order by the coefficient rearranging unit in the high frequency component removing unit is replaced with zero when not encoded by the encoding selection signal. The orthogonal transform quantized coefficient is the variable length coding unit 1.
Variable length coding is performed at 5, and the data is written in the transmission buffer memory unit 16. The encoded data written in the transmission buffer memory unit 16 is output from the output terminal 21 at a constant transmission rate. Further, the orthogonal transform quantized coefficient is inversely quantized by the inverse quantization unit 17, and then inversely orthogonally transformed by the inverse orthogonal transform unit 19. The reproduction image calculation unit 20 adds the data including the quantization error output from the inverse orthogonal transformation unit 19 and the reproduction image calculation data to calculate the reproduction image of the coding block.

[0014]

As described above, according to the configuration of the present invention, by removing the high frequency component of the block determined as the flat region, the influence of the flicker on the screen due to the high frequency noise of the flat region is suppressed. The increase in quantity can be suppressed. Further, as a result, the code amount depending on the image quality can be suppressed, and as a result, frame skips are reduced, and an image with smooth motion can be obtained.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a moving image coding apparatus of the present invention.

FIG. 2 is a block diagram showing a conventional configuration example.

FIG. 3 is an example of an orthogonal transform coefficient rearranging unit.

FIG. 4 is a diagram showing an embodiment of generation of a coding selection signal and removal of high frequency components.

[Explanation of symbols]

 1 input terminal 2 motion vector detection unit 3 frame memory unit 4 motion compensation prediction unit 5 coding method determination unit 6 intra-loop filter unit 7 prediction error calculation unit 8 switch unit 9 orthogonal transformation unit 10 coefficient rearranging unit 11 quantization unit 12 Average variance calculation unit 13 Coding coefficient selection unit 14 High frequency component removal unit 15 Variable length coding unit 16 Transmission buffer memory unit 17 Inverse quantization unit 18 Inverse rearrangement unit 19 Inverse orthogonal transformation unit 20 Reproduced image calculation unit 21 Output terminal

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical indication H04N 7/133 Z

Claims (1)

[Claims] 1. After digitizing each frame of an input moving image signal, orthogonal transformation and quantization are performed in each block dividing a screen, and the quantized digital data is variable length coded and transmitted. In a moving picture coding device for transmitting via a buffer, a prediction means for predicting a current image from a temporally previous image between two temporally consecutive frames to obtain a prediction value, and the prediction from the current image. Means for subtracting a value to generate a prediction error signal, coding method determination means for determining whether each block is interframe-coded or intraframe-coded, and motion vector detection means for calculating a motion vector , A motion compensating means for compensating for motion using a motion vector, and an average variance calculation for calculating an average and a variance of sums within a search range of absolute difference values obtained when the motion vector is calculated. Means, coefficient rearranging means for rearranging the coefficients in order of frequency in order to increase the number of runs of zeros with respect to the orthogonal transformation coefficient, and means for replacing the coefficient rearranged by the average and the variance with 0. A moving picture coding apparatus having.