JP3004799B2 - Image coding method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture signal encoding method and apparatus. Specifically, it is an object of the present invention to provide an image encoding method and apparatus for removing a flicker which is likely to occur when a moving image signal is transmitted by interpolation prediction encoding.

[0002]

2. Description of the Related Art A moving image signal has a huge amount of information. For this purpose, various methods have been conventionally used for efficiently encoding a moving image signal for efficient image transmission. By using the correlation between pictures (frames or fields) of an image signal,
There is inter-picture prediction coding for predicting a current picture from a previous picture. Image signal of current picture and 1
A difference value for each pixel from the image signal of the immediately preceding picture is obtained as prediction error data, and only the obtained prediction error data is encoded and transmitted. As a result, the information amount of the image to be encoded and transmitted is reduced.

[0003] However, although inter-picture prediction coding is effective in a still region, large prediction error data is generated in a moving region, and the picture quality is degraded. The means used to correct this is motion compensated inter-picture prediction coding. In motion-compensated inter-picture prediction coding, before calculating prediction error data between pictures, a motion vector, which is the amount of movement between the current picture to be coded and the immediately preceding decoded picture, is detected. I do. When a motion vector is obtained, prediction error data is calculated for the current picture to be coded at a position shifted according to the motion vector in the immediately preceding decoded picture. The obtained prediction error data is encoded and transmitted to the receiving side together with the motion vector.

If the prediction error data between pictures corrected using a motion vector is encoded and transmitted, the amount of transmission code is reduced as compared with inter-frame encoding without motion compensation.

[0005] However, even when the processing block of the current screen is completely changed and there is no correlation between the pictures, if the inter-picture prediction coding is performed, the prediction error data becomes larger than the data of the current picture only. In some cases. In that case, it is desirable to use intra-picture encoding.

[0006] In the moving image coding for storage media, a type is provided on the screen, and the description will be made with reference to FIG. Here, processing is performed in block units. I picture (Intra pict
ure), intra-picture encoding is performed on all blocks included in the picture. P picture (Predictive
picture), inter-picture prediction coding and intra-picture coding are selected for each block. Hereinafter, the terms I and P pictures mean I pictures or P pictures.

In a B picture (Bidirectional picture), not only the forward (past) I and P pictures are used for prediction (including motion compensation), but the backward (future) I and P pictures are used.
Pictures can also be used for prediction. This is done by changing the processing order of the input images so that decoded images of future I and P pictures are predicted (the processing of B pictures is
P picture). In a B picture, four types of predictions are selected for each block: intra-picture encoding (the prediction signal is 0), forward prediction encoding, backward prediction encoding, and interpolation prediction encoding.

When interpolation prediction coding for detecting prediction error data using the front and rear I and P pictures is selected, the two front and rear I and P pictures for one block in a B picture are selected. Two pieces of motion vector information are used with reference to a P picture. The values of these two pieces of motion vector information are:
It is determined in the I and P pictures that are the reference pictures at the front and rear. The interpolated prediction signal is formed by averaging the forward prediction signal and the backward prediction signal.

[0009] In general, in the above-described video coding apparatus for storage media, a B picture is compared with an I or P picture.
The image is blurred and reproduced.

This is because the B picture is predicted with reference to both the forward and backward I and P pictures, so that the prediction error is small, and the B picture is an image which is not used for the next prediction. This is because, because of the type, only a smaller code amount is generally assigned to the B picture than to the I and P pictures. In addition, since a picture is obtained from a B picture by a prediction signal obtained by averaging prediction signals from both the front and rear I and P pictures, a fine portion (texture) of the picture is lost and the picture is easily blurred.

Such an image blur phenomenon means that the AC amplitude of the image signal is reduced. It is felt that the average level of luminance has decreased because the input signal is subjected to a non-linear characteristic called a gamma characteristic in consideration of the non-linear characteristic of a display device typified by a cathode ray tube by reducing the AC amplitude.

The relationship between the input signal level Y, which is an image signal, and the luminance I on the display will be described with reference to FIG. Here, for the sake of display, the curvature of the curve shown by the thick line representing the inverse gamma characteristic is expressed large.

For example, when the AC amplitude of a clear vertical stripe image signal is large, and the minimum and maximum values of the input signal level are represented by Y _A1 and Y _B1 , points A 1 and A 1 on the curve are obtained.
B1 corresponds, and the average luminance perceived visually in this case is the luminance I _C1 of _C1 at the midpoint between points A1 and B1. However, when the image of the vertical stripes is blurred, the AC amplitude of the image signal becomes small, and the minimum and maximum values of the input signal level are represented by Y _A2 and Y _B2 , and points A2 and B2 on the curve. Corresponds. Then, the average luminance at this time becomes the luminance I _C2 of C2 at the midpoint between points A2 and B2, and the average luminance level is perceived as being lower by the difference between I _C1 and I _C2 . That is, the blurred B picture is perceived as having lower luminance than clear I and P pictures.

[0014]

Since the preceding and succeeding I and P pictures are clear, and the B picture is perceived with a reduced brightness due to blurring, the B picture and the I and P pictures are alternately arranged in time. And a problem to be solved, which is perceived as flicker having a period of I and P pictures.

[0015]

Means are provided for obtaining a difference between an input signal which is an image signal and a prediction signal, and adding a correction signal for correcting luminance on a display (CRT) to obtain a prediction error signal. .

[0016]

By such means, the input signal levels Y _A2 and Y _B2 of the B picture which is the blurred current picture in FIG. 5 are shifted to the right, and the corresponding points A2 and B2 on the curve and the intermediate point C thereof
2 is shifted to the right by the correction signal so that the luminance level I _C2 of the intermediate point C2 matches the luminance level I _C1 . As a result, flicker can be reduced and a more natural image can be reproduced.

[0017]

FIG. 1 shows an embodiment of the present invention.

The B picture which is the current image signal is supplied to the input terminal 4
It is input from 0 and applied to the adder 20 and the correction circuit 21. The correction circuit 21 receives the image signal of the B picture and the prediction signal 45 applied to the input terminal 40, and outputs a correction signal 50 for luminance correction described later.

The adder 20 calculates the sum of the signal level of each pixel included in the block to be processed in the B picture and the correction signal applied via the terminal a of the changeover switch 36, and then calculates the sum. The prediction error signal 44 is obtained by subtracting the prediction signal 45.

The prediction error signal 44 is orthogonally transformed by the transformer 22 that performs orthogonal transformation, and is quantized by the quantizer 23 to output coded data 41. Encoded data 4
1 is transmitted to the receiving side and is inversely quantized by the inverse quantizer 24, and is inversely orthogonally transformed by the inverse transformer 25 to obtain a quantization error signal 47.

The adder 26 is supplied with the quantization error signal 47 and the prediction signal 45, and adds the decoded signal 51 to the decoded signal 51. The decoded signal 51 is supplied to the picture memory 30 or 32 via the terminal d or e of the changeover switch 27. Applied and stored. The image signal of the B picture is applied to the input terminal 40, and the changeover switch 27 is connected to the terminal c when processing the B picture.
The switch 27 is connected to the terminal d or e when the image signal of the P picture is being applied and the front or rear I or P picture is being processed. Therefore, the decoded signal 51 at the time of processing the front or rear I or P picture is stored in the picture memories 30 and 32. Motion compensation circuit 3
1 and 33, using the motion vectors at the forward or backward I, P stored in the picture memory 30 or 32
The picture signal of the motion-compensated block is read out from the picture to obtain a motion-compensated signal 48 or 49, which is connected to the interpolation circuit 34 and the terminals f and h of the changeover switch 35.
Is applied to

The interpolation circuit 34, which has received the motion-compensated signals 48 and 49 from the front and rear I and P pictures, calculates an interpolation prediction signal and outputs an interpolation prediction signal 46, which is a changeover switch. 35 is applied to the terminal g.

When the terminal f or h is selected by the changeover switch 35, a motion-compensated signal from the front or rear I or P picture is selected, and the selected signal is used as the prediction signal 4.
5 will be output. When the terminal g is selected, the interpolation prediction signal 46 is output as the prediction signal 45. When the terminal i is selected, a zero (0) value 43 is output as the prediction signal 45. Switching of the changeover switch 35 is performed by an external control signal (not shown) by selecting an optimum mode for each block in the image.

The terminal a is selected by the changeover switch 36 regardless of the selection of the changeover switch 35 at the time of processing a B picture. When processing I and P pictures, the terminal b of the changeover switch 36 is selected. The control of the changeover switch 36 is performed by a control signal (not shown).

When processing the front or rear I and P pictures, either the terminal e or d of the changeover switch 27 is alternately selected. This will be described with reference to FIG.

In FIG. 2, B pictures are B1, B2,
B3 ..., the I and P pictures are I1, I2, I
, And shows a case where B pictures and I pictures (I, P pictures) are alternately arranged.

It is assumed that when the I1 picture is processed, the terminal d of the changeover switch 27 is selected, and the decoded signal 51 of the I1 picture which is the picture preceding the B1 picture is stored in the picture memory 30. Next, the changeover switch 27 is switched to the terminal e, and the I2 picture which is the picture behind the B1 picture is encoded. At this time,
A signal is transmitted from the picture memory 30 through the motion compensation circuit 31, and the changeover switch 35 selects the terminal h. The decoded signal 51 of the I2 picture is stored in the picture memory 32. Since the screens used for the preceding and following predictions are prepared, the B1 picture is encoded next. During B picture processing, the changeover switch 27 is connected to the terminal c. To predict the B1 picture, the changeover switch 35 can use any of the terminals i, h, g, and f.

Next, the changeover switch 27 is switched to the terminal d, and the I3 picture, which is the picture behind the B2 picture, is coded.
Is selected. The decoded signal 51 is stored in the picture memory 30, and the decoded signal 51 is already stored in the picture memory 32.
Is used as the picture preceding the B2 picture, so that the prediction signal 45 of the next coded picture B2 picture can be obtained from the changeover switch 35.

Next, the operation of the correction circuit 21 will be described with reference to FIG. FIG. 3 shows the same curve as the luminance characteristic shown in FIG. 5, and points A1, B1, C1 and A shown in FIG.
2, B2 and C2 are the same as those already described in the section of the prior art. That is, according to the conventional interpolation predictive coding, the average luminance of the front or rear I or P picture which is a clear image is indicated by I _C1 , and the average luminance of the B picture which is a blurred image is I _C2 . To be shown,
Since the B picture is perceived at a lower luminance than the I and P pictures, flicker has occurred.

[0030] Therefore, in the correction circuit 21, an input signal level Y _A2 of FIG. 3 in Y _A3, Y _B2 are moved parallel to the Y _B3, luminance I _C3 luminance I midpoint C3 of both points A3, B3 _The input signal level is shifted by Y _A3 -Y _A2 = Y _B3 -Y _B2 in order to generate a correction signal 50 for making it equal to _{C 1} and to realize I _C1 = I _C3 .

For this reason, in the correction circuit 21, the input terminal 4
The following operation is performed upon receiving the image signal of the B picture from 0 and the prediction signal 45.

It is assumed that the input signal level of the image signal from the input terminal 40 is Y, the luminance on the display (for example, a cathode ray tube) is I, and the inverse gamma characteristic is I = f (Y). It is assumed that an image signal is processed in units of blocks in a picture, an input signal level in a block is Y _ij, and i and j indicate spatial positions in the block. The number of pixels to be processed in the block is N, the signal level of the prediction signal 45 (FIG. 1) is Y _2ij , and the level of the correction signal 50 is α.
Therefore, an average of the luminance corresponding to the input signal Y _ij in the block included in the current image signal applied to the input terminal 40 and the display corresponding to the signal level Y _2ij of the predicted signal 45 after the luminance correction (FIG. 1) (Not shown) to reduce the flicker so that the average of the above luminances coincides, (1 / N) Σf (Y _ij ) = (1 / N) Σf (Y _2ij + α) ( 1) What is necessary is to make it become. Here, Σ represents a cumulative sum for i and j. The same applies hereinafter.

The inverse gamma characteristic can be approximately expressed by I = kY ² (2). Here, k is a constant, and 2 of the exponent is a value of 2.2 to 2.8. However, when 2 is adopted, implementation (operation) is easy.

From equations (1) and (2), _{ΣY ij} ² = Σ (Y _2ij + α) ² (3) is obtained. From equation (3), α ² + 2αΣY _2ij + _{ΣY 2ij} ² −ΣY _ij ² = 0 (4) is obtained, and from equation (4), α = −ΣY _2ij + β ^1/2 (5) is obtained. Here, β = ( _{ΣY 2ij} ) ² + ΣY _ij ² −ΣY _2ij ² .

Further, when the condition of α << _{2Y 2ij} is used, α = (ΣY _ij ² -ΣY _2ij ² ) (2ΣY _2ij ) ^-1 (6) is obtained.

The value of α shown in the equation (5) or (6) represents the level of the correction signal 50. The α thus obtained is added in advance as a correction signal 50 to the image signal of the B picture from the input terminal 40 at the time of interpolation prediction encoding of the B picture, so that the blurred B picture and the flicker caused by the gamma characteristic are obtained. Can be reduced.

[0037]

As is apparent from the above description, according to the present invention, the display apparatus can be used for an image signal using interpolation prediction coding in which a clear image and a blurry image are alternately reproduced. It is possible to reduce flicker caused by a gamma characteristic in consideration of a non-linear characteristic. Therefore, the effect of the present invention is extremely large.

[Brief description of the drawings]

FIG. 1 is a circuit diagram showing an embodiment of the present invention.

FIG. 2 is a picture position diagram for explaining a switching operation of two picture memories which are components of FIG. 1;

FIG. 3 is a luminance characteristic diagram with respect to an input signal level for describing an operation of performing luminance correction in a correction circuit which is a component of FIG. 1;

FIG. 4 is a picture position diagram for explaining a conventional operation of detecting a motion vector from front and rear pictures.

FIG. 5 is a luminance characteristic diagram showing luminance with respect to an input signal at the time of conventional interpolation prediction encoding.

[Explanation of symbols]

 Reference Signs List 20 adder 21 correction circuit 22 converter 23 quantizer 24 inverse quantizer 25 inverse transformer 26 adder 27 changeover switch 30, 32 picture memory 31, 33 motion compensation circuit 34 interpolation circuit 35, 36 changeover switch 40 Input terminal 41 Encoded data 43 Zero value 44 Prediction error signal 45 Prediction signal 46 Interpolation prediction signal 47 Quantization error signal 48, 49 Motion compensated signal 50 Correction signal 51 Decoded signal 100, 101 Picture 102, 103 Block 104 Search area m, n picture interval

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-58-215186 (JP, A) JP-A-2-164187 (JP, A) JP-A-4-888 (JP, A) (58) Survey Field (Int.Cl. ⁷ , DB name) H04N ^7/ 24-7/68 H04N 5/14-5/217

Claims (2)

(57) [Claims] 1. A difference signal between an image signal (40) of a current picture and a prediction signal (45) obtained from a picture before and after the current picture, and a correction signal (50) for correcting the luminance of the current picture is obtained. In addition (20), a prediction error signal (44) is obtained, and from the prediction signal and the image signal of the current picture, the luminance of the prediction signal on the display is changed by the luminance of the image signal of the current picture on the display. An image coding method for obtaining the correction signal for correcting the prediction signal so as to approach luminance. 2. A correction signal (50) for correcting the luminance of the current picture by taking the difference between the image signal (40) of the current picture and the prediction signal (45) obtained from the preceding and following pictures. In addition, first accumulating means (20) for obtaining a prediction error signal (44), and memory means (3) for temporarily storing each of the decoded signals obtained from the forward and backward pictures.
0, 32), and a motion vector detected from the front and rear pictures stored in the memory means, and a motion compensated signal (4
8, 49) for outputting the motion compensating means (31, 3).
3), interpolation means (46) for outputting an interpolation prediction signal (46) from the motion-compensated signal obtained from the front and rear pictures, and the prediction signal and the image signal of the current picture. Receiving
A correction unit (21) for outputting the correction signal for correcting the prediction signal so that the luminance of the prediction signal on the display approaches the luminance of the display of the current picture.