JPH0771251B2 - Image compression recording system - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression recording system in which prediction encoding of a moving image signal is devised in consideration of reverse reproduction.

[Prior Art] Attempts have been made to provide an interactive video source for education or entertainment by writing image data obtained by performing color image compression and encoding on a CD-ROM together with audio signals. For example, when a color image for one hour is stored in a CD-ROM or the like, which has a limited recording capacity and is difficult to increase in capacity, it is usually necessary to compress an average of up to 5 Kbytes per image of one frame. In that case, the compression ratio reaches 1/120. For this reason, a normal color image requires a transfer rate of 600 Kbytes per frame and a transfer rate of about 18 Mbytes / second, and when digital image data is simply stored in a CD-ROM with a storage capacity of about 540 Mbytes, Play time is short
It takes about 30 seconds, which makes it practically unusable.

However, by introducing a high-efficiency coding method that is mainly composed of prediction error recording and unequal length code recording by the predictive coding shown below, it is possible to record normal audio data without increasing the rotation speed of the playback device. Similar to the CD, a color image CD-ROM having a display time of about 1 hour has been proposed. The high-efficiency coding method was originally developed for the purpose of reducing the average number of bits per pixel of an image, and in particular, predictive coding using interframe DPCM (differential pulse code modulation). It is said that the () method is suitable for processing a television signal having a high correlation between frames and a small frame difference signal.

In this predictive coding method, a predictive value predicted using the data of one frame before is subtracted from the image data of pixels in the XY two-dimensional image plane, and the difference is encoded as predictive error data to compress the image. It is a method. In that case, since the prediction error data is approximated by a Laplace distribution, in the conventional compressed image data recording / reproducing system 1 shown in FIG. 3, the quantizing circuit 3 in the encoder 2 used in the recording system has a logarithmic logarithm. A non-linear quantization circuit by 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 quantizing circuit 3 is converted into an unequal length code in the code converting circuit 5 and recorded in the CD-ROM 6, while the local decoder 7
Is fed back to the subtractor 4 via. This local decoder 7 is
The output of the inverse quantization circuit 8 that performs the inverse processing of the quantization, that is, the conversion from the level number to the level value, is positively fed back to the adder 9 through the prediction circuit 10, and the locally decoded signal obtained from the adder 9 is subtracted. 4 is used.

By the way, the prediction circuit 10 is based on inter-frame prediction, but has a configuration that complements by inter-frame prediction. That is, for each image unit of a plurality of frames forming a moving image, each frame is divided into strips of m vertical stages, for example, if it is the i + km (k = 0,1,2 ...) th frame, Of the image data V (1, i + km), V (2, i + km), ..V (m, i + km) for each split,
Regarding the image data V (i, i + km) of the i-th split,
The split refresh method is used to improve image quality by performing predictive coding based on intraframe prediction.
The split refreshed by intra-frame prediction is an encoding control circuit that controls the prediction mode of the prediction circuit 10.
According to 11, it is circulated for each frame, and it is considered that the refresh effect is made uniform over the entire image. Therefore, the prediction error data recorded on the CD-ROM 6 is the prediction error data E (i, j) by inter-frame prediction when j ≠ i + km with respect to the image data V (i, j) of the split number i and the frame number j. = V (i, j) -V (i, j) is generated, and at j = i + km, prediction error data E (i, j) '= F [V (i, j)] by intra-frame prediction is generated. It However, V (i, j) is a predicted value of V (i, j), that is, image data V (i, j-1) one frame before, and F represents an intraframe coding function. It should be noted that the dash code indicates that it is based on intra-frame prediction.
FIG. 4 exemplifies a prediction error data string when the number of splits m is 4.

On the other hand, the prediction error data E (i.
The decoder 12 for decoding j) or E (i, j) 'reversely transforms the prediction error data code-transformed by unequal-length coding and restores it to the equal-length code. An inverse quantization circuit 14 that inversely quantizes the output of the inverse transformation circuit 13, and an inverse quantization circuit 14
It comprises an adder 15 and a prediction circuit 16 which form the image data V (i, j) from the output prediction error data. The adder 15 includes a prediction circuit 16 based on the output of the dequantization circuit 14 and its own output.
Output of the prediction circuit 16 is cyclically added to the prediction error data to obtain the image data V
(Ij) is played. A decoding control circuit 17 is connected to the prediction circuit 16 and executes prediction mode control that matches the prediction mode used in the process of generating prediction error data. The image data V (i,
j) is obtained by adding the prediction error data E (i, j) to the prediction value V (i, j) when j ≠ i + km, and V (i, j) = V (i, j) + E (i, j), and for j = i + km, we have V (i, j) = G [E (i, j) ']. However, G represents an intraframe decoding function.

[Problems to be Solved by the Invention] The conventional compressed image data recording / reproducing system 1 is a CD
Since the ROM 6 is reproduced according to the recording order, that is, the so-called normal rotation reproduction is assumed, for example, the prediction circuit in the decoder 12 is used during so-called reverse reproduction when reproducing in reverse to the recording order of the prediction error data. Since there is no past data that 16 needs as an initial value, regular reverse playback cannot be done not only in the middle of the image but also from the end image, and even if reverse playback is attempted, it will be There is a problem that a meaningless image which is separated from each other is reproduced, and the image signal cannot be reproduced in reverse by only repeating subtraction between the frames or within the frame.

[Means for Solving the Problems] The present invention solves the above problems and interlaces a moving image signal composed of a plurality of image units with intra-image unit predictive coding for each image unit. A compression recording system for an image that is compressed by inter-image-unit predictive coding and is recorded on a compressed image data recording medium as prediction error data, wherein the image portion compressed by the intra-image-unit predictive encoding is said image first. Intra-unit coding is performed, then the inter-image unit coding is also performed using an image signal obtained by decoding the intra-unit coding, and an encoder is provided to generate coded data by both of these codings. It is a feature.

[Operation] According to the present invention, a moving image signal including a plurality of image units is
In the image part compressed by inter-image predictive coding while compressing by inter-image predictive coding while interweaving intra-image unit predictive coding for each image unit and recording as prediction error data in the compressed image data recording medium By generating prediction error data to which data necessary for accurate image reproduction during reverse rotation reproduction is added, reverse reproduction can be performed from any image unit.

[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 image compression recording system of the present invention is applied, and FIG. 2 is shown in FIG.
It is a figure which shows the content of the prediction error data recorded on CD-ROM.

In FIG. 1, a compressed image data recording / reproducing system 20 is different from the conventional one in the configuration of an image compression recording system including an encoder 21. The encoder 21 anticipates the reverse reproduction of the CD-ROM 6, and improves on the conventional recording method in which recording is performed in consideration of only the normal reproduction. That is, for the image portion compressed by intraframe predictive coding, the prediction circuit 10 and the code conversion circuit 5 are generated so that prediction error data to which data necessary for forming an initial image at the time of reciprocal point reproduction is added is generated.
Is controlled by the encoding control circuit 22.

Here, there is a case where compressed image data is recorded on the CD-ROM 6 using the encoder 21. In this case, as described above, the image data V (i, i, k) relating to the i-th split (j ≠ i + km) of the j-th frame is recorded. For j), the inter-frame prediction mode is set for the prediction circuit 10.

On the other hand, for the image data V (i, j) related to the split refreshed by intra-frame prediction, the prediction circuit 1
The intra-frame prediction mode is set for 0, and the intra-frame prediction calculation F [V (i, j)] performs prediction error data E (i,
j) ′ is obtained. However, the prediction error data E (i, j) 'obtained here is not immediately recorded, and the encoding control circuit 22
However, first, the image data V (i, j-1) of the previous frame is subtracted from the locally decoded output of the prediction error data E (i, j) ', that is, the intraframe prediction value G [E (i, j)']. Record the resulting data E (i, j). Then, after recording the data E (i, j), the intra-frame prediction error data E (i, j) 'is continuously recorded. That is, as shown in FIG. 2, the intra-frame prediction error data E (i, j) 'includes the data E necessary for forming the next image matching the first image at the time of reverse reproduction.
(I, j) is always prefixed.

On the other hand, also in the decoder 31, the dequantization circuit 13 and the prediction circuit 15 are controlled by the decoding control circuit 32 so as to cope with the reverse reproduction of the CD-ROM 6, and the reverse of the recording order is performed during the reverse reproduction at the i + km frame. For the intra-frame prediction error data E (i, j) ′ read out first by the above, a prediction value G [E (i, j) ′] using intra-frame correlation is generated, and the data E (i, j, The prediction error data E (i-1, j), E (i-2, j) ... Read after j) are configured to generate a prediction value using inter-frame correlation.
Further, in order to eliminate the difference in tense between the decoding by intra-frame prediction and the decoding by inter-frame prediction, a delay circuit 33 having a delay time of one frame period is interposed between the inverse quantization circuit 14 and the adder 15. The decoding control circuit 32 switches the changeover switch 34 in the intra-frame prediction mode to delay the output of the inverse quantization circuit 14 for one frame period. Furthermore, when reversing the recording order and reproducing, the prediction error data E supplied from the CD-ROM 6
The polarity of (i, j) must be inverted as appropriate, so that the decoding control circuit 32 uses the inverse quantum except for the prediction error data E (i, j) 'regarding the split read at the beginning of the reverse reproduction. The output of the digitizing circuit 13 is designed to have a negative sign.

By the way, when the CD-ROM 6 is reversely reproduced, the decoder 31 stops decoding until the first intra-frame prediction error data E (i, j) ′ is supplied, and the first intra-frame prediction error data E (i, j, When j) 'is supplied, the decoding control circuit 32 switches the prediction mode of the prediction circuit 16 to the intra-frame prediction mode, and substantial decoding is started. In this case, the prediction error data E (i, j) ′ relating to the first intraframe predictive-coded split that has been read out is directly subjected to intraframe decoding without adding a negative sign, and the encoding process is performed. The same image data G [E (i, j) '] as the locally decoded output obtained in
Is reproduced, and this forms the first image in reverse reproduction.

Since there is a tense difference for one frame between the head image G [E (i, j) '] and the image reversely reproduced subsequently thereto, as described above, the intra-frame prediction is performed. In the decoding based on the above, the changeover switch 34 is switched to the delay circuit 33 side, and the tense is unified by the signal delay of one frame period.

On the other hand, from the data E (i, j) sent following the prediction error data E (i, j) ', the output of the inverse quantization circuit 13 is given a negative sign, and decoding by interframe prediction is performed. Be seen. Therefore, G [E (i, j) '] is followed by G [E (i, j)'].
(I, j) ']-E (i, j) or V (i, j-1) is reproduced. Then, following V (i, j-1), V (i-1, j-1), V
As in (i-2, j-1) ..., decoding based on sequential interframe prediction is performed, and similar decoding is performed on other subsequent frames.

In this way, the compressed image data recording / playback system 20
A video signal consisting of multiple frames is compressed by interframe predictive coding while interlacing intraframe predictive coding for each frame, and is recorded in CD-ROM 6 as prediction error data, and by intraframe predictive coding. In the compressed image portion V (i, j), data E (i, j), that is, G, which is indispensable for accurate image reproduction during reverse reproduction.
Since the prediction error data E (i, j) 'with [E (i, j)']-V (i-1, j) added is generated, the frame read first during reverse playback is read. Intra prediction error data E
By forming an initial image required for decoding from (i, j) ′ and subtracting the additional data E (i, j) from this initial image,
The next image can be reproduced, and the image is reproduced while sequentially subtracting the prediction error data, so that the backward decoding of the prediction error data based on the prediction using the correlation between frames can be accurately executed. You can Furthermore, since the prediction error data encoded by intra-frame prediction is inserted for each frame, a series of moving images including a plurality of frames can be reproduced in reverse not only at the end but also from any frame. .

In the above embodiment, the compressed image data recording medium is not limited to the CD-ROM 6 and may be another type 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 image units, 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, when performing orthogonal transform coding, an orthogonal transform circuit is required in the preceding stage of the quantizing circuit 3 in the encoder 21, and an inverse orthogonal transform circuit is required in the subsequent stage of the inverse quantizing circuit 8, and the decoder 31 An inverse orthogonal transform circuit is required after the inverse quantization circuit 14 in the inside. 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 21 and the decoder 31 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 21, and a vector index circuit for inverse vector quantization is provided in the latter stage of the dequantization circuit 8. A vector-index inverse transform circuit is required for each, and a vector-index inverse transform circuit for inverse vector quantization is required after the inverse quantization circuit 14 in the decoder 31. 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 21 and the decoder 31 remains unchanged.

[Effect of the Invention] As described above, the present invention compresses a moving image signal composed of a plurality of image units by inter-image unit predictive coding while interlacing intra-image unit predictive coding for each image unit. A structure for recording prediction error data on a compressed image data recording medium and generating prediction error data to which data necessary for accurate image reproduction at reverse reproduction is added in an image portion compressed by intra-image predictive coding Therefore, during reverse playback, the next image can be played back by forming the initial image required for decoding from the initially read intra-image prediction error data and subtracting the additional data from this initial image. In addition, the backward decoding of the prediction error data based on the prediction using the correlation between the image units is performed correctly by reproducing the image while further subtracting the prediction error data. Prediction error data that can be accurately executed, and is encoded by intra-image prediction, is inserted for each image unit, a series of moving images consisting of a plurality of image units, not limited to the end, It has an excellent effect such that reverse reproduction can be performed from any image.

[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 image compression recording system of the present invention is applied, and FIG. 2 is recorded on the CD-ROM shown in FIG. Figure 3 shows the contents of the prediction error data
FIG. 4 is a system configuration diagram showing an example of a conventional compressed image data recording / reproducing system, and FIG. 4 is a diagram showing the content of prediction error data recorded on the CD-ROM shown in FIG. 3 ... Quantization circuit, 4 ... Subtractor, 5 ... Code conversion circuit, 6 ...
... CD-ROM, 7 ... Local decoder, 10 ... Prediction circuit, 20 ... Compressed image data recording / reproducing system, 21 ... Encoder, 22 ...
Encoding control circuit.

 ─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Mutsumi Ota 5-33-1 Shiba, Minato-ku, Tokyo Inside NEC Corporation (72) Hideto Kunihiro Inc. 3-5 Miyahara, Yodogawa-ku, Osaka-shi, Osaka No. 24 in NEC Home Electronics Co., Ltd. (56) Reference JP-A-63-125071 (JP, A)

Claims (1)

[Claims] 1. A moving image signal composed of a plurality of image units
A compression recording system for an image, which is compressed by inter-image unit coding while interweaving intra-image unit encoding for each image unit, and is recorded on a compressed image data recording medium as prediction error data. The image portion to be compressed is first subjected to the intra-image unit encoding, and then the inter-image unit encoding is also performed using the image signal obtained by decoding this, and the encoded data by both of these encodings is obtained. An image compression recording system provided with an encoder for generation.