JPH0642737B2 - Video coding system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention interlaces intra-predictive coding of a moving picture signal almost regularly with intra-predictive coding of inter-picture,
The present invention relates to a moving picture coding system capable of realizing a quick reproduction signal from an arbitrary time point in either forward or reverse directions.

[Prior Art] The high-efficiency encoding method used for compression recording of color moving images is effective in reducing the average number of bits per pixel of an image, and particularly when one frame of a television signal is taken as a unit image The predictive coding method using DPCM (Differential Pulse Code Modulation) is said to be suitable for processing a television signal having a small frame difference signal because of high inter-frame correlation. The predictive coding method compresses by subtracting a predicted value predicted using data of one frame before from image data regarding pixels in an XY two-dimensional image plane and coding the difference as prediction error data. It is a method.

Since the prediction error data can be approximated by a substantially positive distribution, the conventional compressed image data recording / recording data shown in FIG.
In the reproducing system 1, as the quantizing circuit 3 in the encoder 2 used in the recording system, a non-linear quantizing circuit by logarithmic compression is used. The encoder 2 supplies the input image data to the quantization circuit 3 via a subtracter 4 that takes the difference from its predicted value. One of the prediction error data converted from the level value to the level number in the quantization circuit 3 is converted into an unequal length code in the code conversion circuit 5 and recorded in the CD-ROM 6, while
It is fed back to the subtracter 4 via the local decoder 7. The local decoder 7 supplies the output of the inverse quantization circuit 8 for inversely converting the quantization, ie, the level number to the level value, to the subtractor 4 via the predictor 9, while the input side of the predictor 9 is supplied. Positive feedback is provided to the adder 10 provided in the.

In the case of this example, the predictor 9 connects the inter-frame predicting circuit 9a and the intra-frame predicting circuit 9b in parallel, and selectively switches one of them by the selector switch 9c.
It is configured to connect to. This is because, when there is a scene change (scene change), the first image after the scene change has no correlation with the image signal of the previous frame,
This is because the intraframe predictive coding using the intraframe correlation of the image has higher coding efficiency than the interframe predictive coding. When taking a series of unit images obtained by dividing continuous unit images as an example, the prediction mode control circuit 11 operates based on the scene change information from the scene change information generation circuit 11a at the time of a scene change, and the changeover switch 9c is turned on. Intra-frame prediction circuit 9b
Switch to the side. Therefore, C is the first time after a scene change.
The prediction error data E (1) ′ recorded in the D-ROM 6 is
By setting the intra-frame prediction mode for the predictor 9, E (1) '= F [V (1)]. However, the dash code indicates that it is based on intra-frame prediction, and F is a coding function for intra-frame predictive coding. Since the inter-frame prediction mode is set to the predictor 9 from the second frame following the first frame, E (i) = V (i) -V (i) = V for the image of the i-th frame. The prediction error data E (i) obtained by the calculation (i) -V (i-1) is the CD
-Recorded in ROM6. However, V (i) is the predicted value of V (i), that is, the image data V (i-1) of the previous frame.

On the other hand, the decoder 12 for decoding the prediction error data E (i) or E (i) 'read from the CD-ROM 6 inversely transforms the prediction error data code-converted by the unequal length coding, A code inverse conversion circuit 13 for returning to the equal-length code and a code inverse conversion circuit 1
The dequantization circuit 14 dequantizes the output of No. 3 and the adder 15 and the predictor 16 that form the image data V (i) from the output prediction error data of the dequantization circuit 14. The output of the inverse quantization circuit 14 is output through the adder 15 while
The image data V (i) is reproduced by the positive feedback to the adder 15 via the predictor 16 and the cyclic addition of the prediction error data and its decoded output. The predictor 16 is connected to a prediction mode control circuit 17 that switches the prediction mode according to the prediction mode used in the process of generating prediction error data.

The image data V decoded by the decoding device 12
(i) is generated as V (i) = G [E (i) ′] when i = 1, and when i ≧ 2, the prediction error data E ( By adding i), it is generated as V (i) = V (i) + E (i). However, G represents an intraframe decoding function.

[Problems to be Solved by the Invention] In the above-described conventional compressed image data recording / reproducing system 1, whether or not there is a scene change is left to the judgment of the operator involved in the production, and the operator makes the frame by his own judgment. It is also possible to execute the inner prediction coding. It is also possible to use a method in which the prediction error data based on inter-frame prediction and the prediction error data based on intra-frame prediction are constantly energy-compared, and the prediction mode with the smaller power is automatically selected. However, in such a method in which intra-frame prediction is switched and used for inter-frame prediction, intra-frame prediction is mostly carried out sporadically. For example,
In an image that looks like a person against a static background,
Since there is no scene change and almost the same video continues for a long time, inter-frame prediction is continuously executed after intra-frame prediction is performed only for the first frame. Then, when reproducing the CD-ROM 6 that contains almost no such intra-frame prediction error data and occupies most of the inter-frame prediction error data, there is no problem when reproducing from the first scene, but it is not the first image but the intermediate reproduction. In such a case, it takes a considerable amount of time until the prediction error data obtained by the intra-frame prediction is obtained next time, and a meaningless image continues to be reproduced during that time. There is a problem that it is impossible to reproduce in the middle when using the inner prediction.

When the encoding technique is used in the recording / reproducing system as in the present invention, a data error at the time of reproduction is corrected on the system side, and it is not necessary to take any error countermeasure unlike the case of using for communication. However, halfway reproduction is possible only when the intra-frame prediction is used here.

In addition, the above-mentioned conventional compressed image data recording / reproducing system 1
Has been configured on the assumption that only so-called normal rotation reproduction is performed, so that the predictor 16 in the decoder 12 is required as an initial value at the time of so-called reverse reproduction, in which reproduction is performed in a manner reverse to the recording order of the prediction error data. Since there is no past data, regular reverse playback cannot be performed not only in the middle of the image but also from the end image, and even if reverse playback is attempted, a meaningless image that is far from the original image However, there is a problem in that the image signal cannot be reproduced in reverse by only repeating subtraction between the frames or within the frame.

The reason why normal reverse reproduction cannot be performed from the end image is that the quantization is usually applied to the prediction error, and the quantization error generated at that time is unavoidable.

[Means for Solving the Problems] The present invention is used in a system that encodes a moving image signal composed of continuous unit images and reproduces the moving image signal from a recorded recording medium. Is a method of encoding by switching between a first encoding method utilizing the above and a second encoding method utilizing the correlation within the unit image, and is obtained by dividing each of the plurality of unit images. The first unit image of the series of unit images is encoded by the second encoding method, and the others are encoded by the first encoding method, at least before encoding the last unit image of the series of unit images. The first unit image of the series of unit images is encoded by the second encoding method.

[Operation] According to the present invention, a moving image signal is encoded by switching the intra-predictive coding to the inter-predictive coding almost regularly, and the initial image forming time for the decoding request from the middle of the scene is set. Shorten. Further, the decoded image for the intra-unit image predictive coding is used to shorten the initial image formation time even in reproduction in either the forward or reverse direction.

[Embodiment] An embodiment of the present invention will be described below with reference to FIGS. FIG. 1 is a system configuration diagram showing an embodiment of a compressed image data recording / reproducing system to which the moving picture coding system of the present invention is applied, and FIG. 2 is a CD-recording system shown in FIG.
It is a figure which shows the content of the prediction error data recorded on ROM.

In FIG. 1, the compressed image data recording / reproducing system 21 is
The configuration of the image compression recording system including the encoder 22 is different from the conventional one. This encoder 22 is used by periodically switching to intraframe prediction for interframe prediction so as to reduce the initial image formation time in response to a decoding request from the middle of the scene, and is devised to record intraframe prediction error data. By doing so, the CD-ROM 6 can be reversely reproduced from the middle of the scene. In addition, quantization is applied to prediction error data obtained by intra-frame prediction and intra-frame prediction.

Reference numeral 23 is a counter circuit connected to the prediction mode control circuit 11, and after the prediction mode control circuit 11 switches the prediction mode from within the frame to between the frames, the counter circuit decrements by 1 from a predetermined count value set in advance according to a constant clock. , The prediction mode switching is instructed when the count value matches zero. Of course, if there is scene change during the subtraction operation, the subtraction from the set count value is restarted again at the time of scene change. Therefore, regardless of the presence or absence of scene change, the prediction mode control circuit 11 will periodically switch to the intra-frame prediction mode within the period T required to subtract the count value. For this reason, a CD in which such intra-frame prediction error data is regularly inserted and recorded.
-When the ROM 6 is reproduced, when the reproduction is performed from the beginning image in the middle, it is possible to obtain the prediction error data by the intra-frame prediction at most within the period T, and the initial image can be reproduced immediately at that time. Become.

By the way, with respect to the image data V (j) of the j-th frame which is the target of predictive coding by intraframe prediction, the prediction error data E (j) ′ is calculated by the intraframe prediction calculation F [V (j)].
Needless to say, is obtained. However, since the prediction error data E (j) 'obtained here is required to correctly decode the next image at the time of reverse reproduction, the locally decoded output of the prediction error data E (j)', that is, the intraframe prediction value G [E (j) '], the data E (j) obtained by subtracting the image data V (j-1) of the previous frame is waited for and then recorded. That is, at the same time as the prediction coding in the intra-frame prediction mode is completed, the prediction mode control circuit 11
Stops the supply of image data and connects to the preceding stage of the subtractor 4 to switch the input of the multiplex circuit 24 to the frame delay circuit 25 side connected to the inter-frame prediction circuit 9a. As a result, interframe coding G [E (j) ']-V (j-1) is executed with the image data V (j-1) output from the frame delay circuit 25 as input, and E (j) Is obtained. Here, since the intra-frame prediction error E (j) ′ is quantized and includes the quantization error, G [E (j) ′] = V (j) does not hold, and in this order Encoding is required.

Then, after the interframe predictive coding is completed, the prediction mode control circuit 11 releases the stop of the image input during this period,
Ordinary interframe coding is restarted, but with respect to the prediction error data E (j) 'and E (j) and the prediction error data E (j + 1) following this, E (j), E ( It is necessary to output j) ′ and E (j + 1) in that order. For this reason, in the embodiment, a line 28 having a frame delay circuit 26 and a 2-frame delay circuit 27 as a bypass is provided at the subsequent stage of the quantizing circuit 3, and these lines are controlled by the predictive mode control circuit 11 to have a 3-input alternative output. Type multiplex circuit 29. That is, when switching to the intra-frame prediction mode, the input of the multiplex circuit 29 is switched from the line 28 to the two-frame delay circuit 27 side, and the prediction error data E (j) 'is delayed by two frame periods. Next, when the intra-frame prediction mode is switched to the inter-frame prediction mode, the input of the multiplex circuit 29 is switched to the line 28 side and the prediction error data E (j) is output as it is. Then, at the same time when the normal interframe prediction is restarted, the multiplex circuit 29
Is switched to the frame delay circuit 26 side, the prediction error data E (j + 1) is delayed by one frame period and output.

Thus, as shown in FIG. 2, the intra-frame prediction error data E (j) 'is always prefixed with the data E (j) necessary for forming the next image that matches the first image at the time of reverse reproduction. Will be done. On the other hand, in response to such processing on the encoder 22 side, the decoder 3 is slightly devised to enable reverse reproduction.
2 is also required. That is, the decoder 32 uses the initial intra-frame prediction error data E read at the beginning of reverse reproduction.
For (j) ', the predicted value G using intra-frame correlation
[E (j) '] is generated, and prediction error data E (j-1), E (j-2), etc., which are read subsequently to the data E (j), are predicted using interframe correlation. It suffices to generate a value, but a circuit that changes the temporal order of unit images in decoding by intra-frame prediction and decoding by inter-frame prediction, or E (j), E that follows the initial prediction error data E (j) '
It is necessary to add a polarity inversion circuit or the like that inverts the polarity of (j-1).

In this way, the compressed image data recording / reproducing system 21
Is able to shorten the initial image formation time for a decoding request from the middle of a scene by coding the moving image signal while interlacing the intraframe predictive coding with the intraframe predictive coding almost regularly. By forcing the intra-frame predictive coding regardless of the mechanical judgment conditions and the intention of the operator, it is possible to respond to an arbitrary arbitrary reproduction request that does not specify a scene in a short time.

In addition, the compressed image data recording / reproducing system 21 uses the data E (j) that is indispensable for reverse reproduction, that is, G [E], in the image portion V (j) compressed by intraframe predictive coding.
(j) ']-V (j-1) is added to the prediction error data E.
Since (j) 'is generated, an initial image required for decoding is formed from the initially read intra-frame prediction error data E (j)' at the time of reverse reproduction, and then the additional data E ( By subtracting j), the next image can be reproduced, and by further subtracting the prediction error data and reproducing the image, the reverse direction of the prediction error data based on the prediction using the correlation between frames The decryption can be performed exactly.

In the above embodiment, the compressed image data recording medium is not limited to the CD-ROM 6, but may be another digitally recordable medium such as a video tape.

Further, in the above description, inter-frame differential encoding is taken as an example of the predictive encoding method using the correlation between unit images, but other methods such as motion-compensated inter-frame prediction can also be used. . Further, it is also possible to perform orthogonal transform coding or vector quantization on the prediction error data generated in these coding processes. For example, in the case of performing orthogonal transform coding, an orthogonal transform circuit is required before the quantizing circuit 3 in the encoder 22, and an inverse orthogonal transform circuit is required after the dequantizing circuit 8. In this case, what is recorded on the recording medium is the prediction error data converted into the conversion coefficient component, but the essence of the encoder 22 remains unchanged. When performing vector quantization, a vector index circuit for vector quantization is provided in the preceding stage of the quantization circuit 3 in the encoder 22, and a vector index circuit for inverse vector quantization is provided in the latter stage of the dequantization circuit 8. A vector / index inverse conversion circuit may be provided. In this case, what is recorded on the recording medium is the vector index converted from the prediction error data, but the essence of the encoder 22 remains unchanged.

[Effect of the Invention] As described above, according to the present invention, a moving image signal is encoded by switching between intra-image predictive encodings almost regularly for use in inter-image predictive encoding. , The initial image formation time for a decoding request from the middle of the scene can be shortened, and the intra-unit image predictive coding is forced regardless of the mechanical judgment conditions or the intention of the operator, so that no scene is specified. It is possible to construct a moving image reproduction system capable of responding to the reproduction request in a short time.

Further, according to the present invention, when reverse reproduction is performed, an initial image required for decoding is formed from the prediction error data in the unit image that is read first, and additional data is subtracted from this initial image to reproduce the next image. By reproducing the image while sequentially subtracting the prediction error data, backward decoding of the prediction error data based on the prediction using the correlation between the unit images can be accurately executed. Produce an effect.

[Brief description of drawings]

FIG. 1 is a system configuration diagram showing an embodiment of a compressed image data recording / reproducing system to which the moving picture coding system of the present invention is applied, and FIG. 2 is recorded on the CD-ROM shown in FIG. FIG. 3 is a diagram showing the contents of prediction error data, and FIG. 3 is a system configuration diagram showing an example of a compressed image data recording / reproducing system to which a conventional moving image coding system is applied. 9a ... Inter-frame prediction circuit, 9b ... In-frame prediction circuit, 11 ... Prediction mode control circuit, 21 ... Compressed image data recording / reproducing circuit, 22 ... Encoder, 23 ... Counter circuit, 24, 29 ... Multiplex circuits, 2
5, 26 ... Frame delay circuit, 27 ... 2 frame delay circuit, 28 ... Line.

Front page continuation (72) Inventor Mutsumi Ota 5-33-1 Shiba, Minato-ku, Tokyo Inside NEC Corporation (72) Hideto Kunihiro 3-5-24 Miyahara, Yodogawa-ku, Osaka, Japan Electric Home Electronics Co., Ltd. (56) Reference JP-A-60-66584 (JP, A)

Claims (3)

[Claims] 1. A first encoding method used in a system for encoding a moving image signal composed of continuous unit images and reproducing the moving image signal from a recorded recording medium, and utilizing the correlation between the unit images. , And a second encoding method that uses the correlation in the unit image by switching, and is used for encoding, and the first unit image of a series of unit images obtained by dividing each unit image is In encoding by the second encoding method and the others by the first encoding method, at least before the final unit image of the series of unit images is encoded, the first unit image of the next sequence of unit images is encoded. Is encoded by the second encoding method. 2. A decoded unit image obtained by encoding and decoding the head unit image of a series of unit images by a second encoding method utilizing correlation within a unit image, and the immediately preceding series of units. The data representing a prediction error obtained by executing the first encoding method utilizing the correlation between unit images with at least the final unit image of the image is added. The described video encoding method. 3. Quantization is applied to data representing a prediction error obtained by applying a first coding method using correlation between unit images or a second coding method using correlation within unit images. The moving picture coding method according to claim 1, wherein the moving picture coding method is provided.