JP2006141045A - Moving picture decoding apparatus and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a moving picture decoding apparatus and method which can restrain the quality of a decoded image from greatly deteriorating even if an error occurs in transmission data and which can efficiently decode the data even if only a part of the data can be decoded due to the error. <P>SOLUTION: The moving picture decoding apparatus includes: a decoding means for decoding image information in each region, the each frame of which is divided into a plurality of regions and coded and decoding a region rearranging table for describing information denoting the rearranging order in the unit of the regions; and a region rearranging means for rearranging the image signals of each region decoded according to the region rearranging order described in the region rearranging table decoded by the decoding means into the image signals in the correct rearranging order. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a moving image code provided in a device / system for transmitting and storing a moving image and a device / system for reproducing such as a TV phone, a TV conference system, a portable information terminal, a digital video disk system, and a digital TV broadcasting system. More particularly, the present invention relates to an apparatus that compresses and encodes an image to a small amount of information and an apparatus that restores the compressed and encoded information and reproduces the image. In particular, even when transmitting / accumulating an encoded image via a medium that is prone to error such as a wireless transmission path, the moving image encoding / decoding is highly error-resistant and can transmit / accumulate a moving image with high quality. The present invention relates to an apparatus and method.

  In recent years, as a technique for compressing and encoding image information into a small amount of information for efficient transmission and storage, individual methods such as motion compensation, discrete cosine transform, subband encoding, pyramid encoding, and some of these methods are used. Various methods such as a combination of these have been developed and proposed. In particular, as an international standard system for compressing and encoding moving images, ISO-MPEG1, MPEG2, ITU-T.H. 261, H.M. 262 "has been standardized. These are all included in a method using motion compensation adaptive prediction discrete cosine transform coding, and details thereof are described in Non-Patent Document 1 and the like. Here, MPEG is the name of an organization (Moving Picture Experts Group) for proceeding with the standardization work for the color moving image storage encoding method, and also as the standard name of the encoding method whose specifications are defined by this organization. It is used.

  Here, a block diagram of FIG. 29 shows a configuration of an encoding apparatus using motion compensated adaptive prediction discrete cosine transform as an example of a conventional moving picture encoding apparatus. The image 331 input to the encoding device is first divided into regions determined by the region divider 301. Then, as shown in FIG. 30A, encoding is performed in block order in the order from the upper left area to the lower right area, and a code string continuous to the frame synchronization signal is formed as shown in FIG. The In the encoder, first, motion compensation adaptive prediction is performed. In FIG. 29, a motion compensation adaptive prediction signal generator 304 detects a motion vector between an input image and an image that has been stored in the frame memory 305 and has already been encoded and locally decoded. Create a compensated prediction signal. A suitable prediction mode is selected from among motion compensation prediction and intra-frame coding (prediction signal = 0) using the input signal as it is for coding, and a corresponding prediction signal 333 is output.

  The subtractor 306 subtracts the prediction signal 333 selected from the input signal 331 and outputs a prediction residual signal 334. The prediction residual signal 334 is subjected to discrete cosine transform (DCT) in a block unit having a constant size in the discrete cosine transformer 307 and further quantized in the quantizer 308. The output from the quantizer is branched into two, and one is multiplexed with a motion vector by a multiplexer 309 and output. The other is inversely quantized by an inverse quantizer 312 and further subjected to inverse discrete cosine transform (inverse DCT) by an inverse discrete cosine transformer 313. The output from the inverse discrete cosine transformer 313 is added to the adaptive prediction signal 333 in the adder 314 and recorded in the frame memory 305.

  Such a conventional video encoding apparatus has the following problems. First, in the variable-length code, if an error occurs in one place, the variable-length code is out of synchronization, and even if subsequent information can be received correctly, it is decoded to an incorrect value. As a result, an error at one location also affects the subsequent decoding, leading to significant image quality degradation. Therefore, conventionally, when an error is detected, processing such as discarding information up to the frame synchronization signal (PSC) of the next frame is performed. However, even if such processing is performed, the amount of information is greatly reduced compared to the original amount of information, so that a great deterioration in image quality occurs.

  As a method for solving such a problem, a high-efficiency video coding scheme considering error tolerance has been proposed by Matsumura and Nakai in the 1994 Image Coding Symposium (PCSJ94) (see Non-Patent Document 2). Although variable-length codes are used in moving picture coding, variable-length codes have a drawback that loss of synchronization occurs when an error occurs. However, the variable-length code also has a feature of automatically recovering synchronization by recovering synchronization automatically by decoding the codeword even when a loss of synchronization occurs due to an error (self-synchronization recovery characteristic). Normally, even if the synchronization is recovered, it is not possible to specify which area corresponds to the information, and the information after the synchronization recovery cannot be used. In the above method, the coding sequence description method is summarized for each information content as shown in FIG. 31 from the conventional arrangement for each region so that the self-synchronization recovery characteristic of the variable length code can be used. It is characterized in that the arrangement of column units is changed. As a result, even if an error occurs, the decoding is performed up to the frame synchronization signal of the next frame as it is, and the decoding value of the prediction residual signal is used in order from the later in time, together with the header information decoded at the front part. Can increase the amount of information available.

  However, this method also has the following problems. In the case of a transmission line with a high error rate in which a plurality of errors exist in one frame of information, usable information is reduced and the effect is reduced. When the area that can be decoded decreases, an area with a high probability of being decoded is biased to a specific location (upper part of the screen) regardless of the nature of the image. This is inefficient because there is an important area at the center of the screen when normal use is considered, and only areas of low importance such as the background are relieved. Furthermore, when trying to predict a region that cannot be decoded from a region that can be decoded, the entire screen must be predicted from a region that is partially biased, which is difficult to predict. Further, when an error occurs in the header information of the first half, even if all the rear prediction residual information can be decoded, only the area where the header information of the first half can be decoded can be decoded. That is, since the prediction residual information of the rear part is information depending on the front header information, information that can be decoded is wasted.

  FIG. 32 is a block diagram showing a schematic configuration of a conventional video decoding device. This decoding apparatus obtains a decoded image signal by performing the reverse operation of the encoding apparatus shown in FIG.

  FIG. 33 is a diagram showing an example of a conventional DCT coefficient scan order, in which a zigzag scan is performed through a path as shown.

FIG. 34 shows an example of a conventional encoding method in which a synchronization signal is inserted in a frame. When the area shown in FIG. 34 (a) is encoded, it is shown in FIG. 34 (b). By inserting a synchronization signal in units of such n block lines (n ≧ 1), the chances of synchronization recovery are increased.
“International standard for multimedia coding” (Hiroshi Yasuda, edited by Maruzen, published in June 1991) "High-efficiency video coding considering error resilience", Matsumura & Nakai, 1994 Image Coding Symposium, 1-1, 1994

  As described above, the conventional moving image encoding device and decoding device have a drawback that the quality of the decoded image is greatly reduced due to loss of synchronization of the variable length code when an error occurs in the code. Further, when only a part of an image can be decoded due to an error, the decodable probability for each region is biased to a specific part regardless of the feature of the image, which is inefficient.

  Even if an error occurs in any part of the code, the present invention suppresses a significant deterioration in the quality of the decoded image, and efficiently decodes even when only a part of the code can be decoded due to the error. An object of the present invention is to provide a moving picture decoding apparatus and method that can be used.

  The moving picture decoding apparatus according to the first configuration of the present invention describes the information indicating the rearrangement order in units of areas while decoding the image information of each area divided into a plurality of areas for each frame and encoded. Decoding means for decoding the area rearrangement table, and an area for rearranging the image signals of each area decoded in accordance with the area rearrangement order described in the area rearrangement table decoded by the decoding means in the correct rearrangement order Reordering means.

  The moving picture decoding apparatus according to the second configuration of the present invention is the video decoding apparatus according to the first configuration, wherein the area rearranging means is configured to decode each area obtained by decoding an image signal input for each frame in units of a plurality of macro blocks. The image signals are rearranged.

  The video decoding device according to the third configuration of the present invention is characterized in that the information indicating the rearrangement order described in the area rearrangement table area is an order number indicating the area rearrangement order.

  In the moving picture decoding apparatus according to the fourth configuration of the present invention, in any one of the first to third configurations, the region rearranging unit may convert the region of the screen where important information is located into the rearrangement order. The area is rearranged at the top of the head.

  Also, in the moving picture decoding apparatus according to the fifth configuration of the present invention, in any of the first to fourth configurations, the decoding means encodes each region image encoded from each divided region. Decoding the first code sequence arranged in the forward direction from the front to the rear in the information in the forward direction from the front to the rear, and decoding the second code sequence arranged in the reverse direction from the rear to the front in the reverse direction from the rear to the front It is characterized by.

  In the moving picture decoding method according to the sixth configuration of the present invention, the image information of each area encoded by being divided into a plurality of areas for each frame is decoded and information indicating the rearrangement order of the area unit is described. A decoding step for decoding the region rearrangement table; and a region rearrangement step for rearranging the image signals of the respective regions decoded in accordance with the region rearrangement order described in the decoded region rearrangement table into a correct rearrangement order. It is characterized by that.

  In the moving image decoding method according to the seventh configuration of the present invention, in the sixth configuration, the first decoding step decodes the image signal input for each frame in units of a plurality of macroblocks. It is characterized by.

  Further, in the moving picture decoding method according to the eighth configuration of the present invention, in any of the sixth and seventh configurations, the information indicating the rearrangement order described in the region rearrangement table region is the region rearrangement order. It is the order number which shows.

  The moving picture decoding method according to the ninth configuration of the present invention is the video decoding method according to any of the sixth to eighth configurations, wherein in the region rearranging step, the region of the screen in which important information is located is the rearrangement order. The area is rearranged at the top of the head.

  The moving picture decoding method according to the tenth configuration of the present invention is the video decoding method according to any one of the sixth to ninth configurations, wherein in the decoding step, the image information of each area encoded from each divided area is used. The first code sequence arranged from the front to the rear is decoded in the forward direction from the front to the rear, and the second code sequence arranged in the reverse direction from the rear to the front is decoded in the reverse direction from the rear to the front. Features.

  As described above, according to the present invention, it is possible to provide a moving picture encoding / decoding apparatus and method in which the quality degradation of a decoded picture caused by a code error occurring during transmission / accumulation is small.

  Dividing the coding information in moving picture coding into a plurality of code information groups, and describing at least one of the code information groups in the forward direction and the other in the reverse direction, an error occurs in one information. However, by decoding from the other information, it is possible to improve the decoding efficiency of the moving image information, and it is possible to prevent the quality of the decoded image from being greatly deteriorated.

  Further, in the case of moving image encoding, when two independent code strings are output from one input image by encoding, the encoding order of each region obtained by dividing the input image is encoded in different orders, respectively. Even when an error occurs in both pieces of information and decoding becomes impossible, the decodable part is not biased to a part of the image, and the entire screen or important part can be preferentially reproduced.

  Embodiments of the present invention will be described below. First, the overall configuration will be described with reference to FIG. 1 which is a conceptual diagram of a moving image encoding / decoding device. The image input is compression encoded by the encoding means 10 and formed into two code strings in a predetermined arrangement. The encoding unit 10 includes an area dividing unit 11 that divides the screen into predetermined areas, an area rearranging unit 12 that rearranges the divided areas on the basis of, for example, importance, and the like. Code string forming means 13 for forming a code string. The code string formed by the encoding means 10 is divided by the code string dividing means 14, and then the code string rearranging means 15 within one frame from the two frame synchronization signals (PSC) sandwiching this frame to the center of the frame. Sorted in both forward and reverse directions. This state is shown in the data frame of FIG.

First embodiment

  FIG. 3 is a block diagram of the first embodiment of the video decoding apparatus according to the present invention. The first embodiment is a moving picture decoding apparatus using a motion compensated adaptive prediction discrete cosine transform coding system.

  Since the motion compensation adaptive prediction discrete cosine transform coding method is described in detail in the aforementioned Non-Patent Document 1, etc., only the outline of the operation will be described, and the difference from the conventional method will be described in detail.

  The image 131 input to the encoding device is divided into predetermined regions by the region divider 101 and then transferred to the region rearranger 102. The region rearranger performs rearrangement based on the information in the region rearrangement table 103 describing the rearrangement order, and transfers it to the motion compensated adaptive predictor 104 for each region. Since the part relating to the rearrangement of the areas is an important part of the present invention, the outline of the block diagram will be described and then described in detail.

  In the motion compensated adaptive predictor 104, the previous frame information called from the frame memory 105 in which the previous frame for which the encoding / decoding process has been completed is stored, and the current frame information divided for each region are obtained. In comparison, from which part of the previous frame the area of the current frame has moved, that is, how much movement has occurred in the area up to the current frame, this is expressed as a motion vector (this A series of operations is called motion compensation).

  The subtractor 106 subtracts each area information of the previous frame that has been motion-compensated from each area information of the current frame, and obtains the difference information (that is, information of the current frame that could not be expressed simply by considering the motion information in the previous frame). Are encoded by the discrete cosine transformer 107 and the quantizer 108. For each region, this encoded information and motion vector information are multiplexed by a multiplexer 109, and the code string is further divided into two by a code string divider 110, and the next code string rearranger 111 is divided. After being rearranged at, it is sent to the transmission line and sent to the receiving side. Details of the code string divider 110 and the code string rearranger 111 will be described later in the same manner as the area rearranger because of an important part of the present invention.

  On the other hand, the encoded information for each region is decoded as a motion compensation error by the inverse quantizer 112 and the inverse discrete cosine transformer 113, and the previous frame information and motion compensation error after motion compensation are added by the adder 114. And the current frame information is reproduced and stored in the frame memory.

  In the conventional motion compensation error encoding method, the region divider 101 divides the region and then encodes the region unit in the order shown in FIG. 30, whereas in the embodiment of the present invention, each region is encoded. After the division, a region rearranger 102 is newly added to rearrange the regions. FIG. 4 is an example showing in what order the divided areas are rearranged. An area rearrangement table 103 storing the coding order of such areas as the order number is prepared, and the area rearranger 102 rearranges the areas while referring to it. In the present invention, the code string rearranged and encoded by the area rearranger 102 is divided into two code strings by the code string divider 110. In the case of FIG. 4, it is divided into a first code string and a second code string corresponding to (a) and (b), respectively.

  Then, in the next code sequence rearrangement unit 111, the first code sequence is arranged in the forward direction from the front to the rear in succession to the frame synchronization signal, and the second code sequence is arranged in the rear of the first code sequence. Rearranged in the reverse direction and sent to the transmission path. FIG. 5 describes the arrangement of the code strings sent to the transmission line in the present invention. As a result, the first code sequence among the code sequences divided into two can be decoded by finding the frame synchronization signal of the current frame, and the second code sequence can be decoded by finding the frame synchronization signal of the next frame. Since the code string is arranged in the reverse direction from the synchronization signal, decoding is possible.

  Therefore, when an error occurs in the middle of decoding, the information up to the frame synchronization signal of the next frame has to be discarded in the past. Even if the information has an error and cannot be decoded, and even if the end portion of the first code string cannot be specified due to loss of synchronization of the variable-length code, the second code string depends only on the frame synchronization signal of the next frame. Therefore, decoding is possible regardless of the first code string. As described above, the present invention has a structure in which the influence of an error occurring in one code string does not reach the decoding of the other code string. This can solve the problem that the rear information depends on the front information, which has been a problem in the conventional coding apparatus of FIG.

  Furthermore, even if one of the information is missing or both of the information can only be decoded halfway, the area encoding order is changed as shown in FIG. Can be avoided. In addition, since the decoded area is uniformly present on the entire screen, it is possible to easily predict the area that cannot be decoded using the information of the decoded area. As a result, the image of the current frame can be prevented from causing a significant deterioration in image quality even if it is not perfect, and can be decoded as an image having a certain high image quality.

  Although the example which combined the 1st and 2nd invention was given as 1st Embodiment, it is not necessary to use the 1st and 2nd invention simultaneously, and it is also possible to implement | achieve each independently. In addition, the first embodiment uses the motion compensated adaptive prediction discrete cosine transform coding method. However, this coding method is not particularly required and can be used for other image coding methods. .

  6 and 7 show some examples of area rearrangement tables. For example, if the input image has important information (movement, face, etc.) at the center of the screen, the information is rearranged so that the center of the screen comes to the top. By performing the rearrangement in this way, even if an error occurs and the synchronization of the variable length code is lost, the probability of decoding an important area can be made higher than before. This is because, in the nature of the variable-length code, when the error occurs, the rear information has a higher probability that it cannot be decoded. In addition to this, an important center part of an image is obtained from the value of a motion vector, etc., and the region is rearranged for each input image by encoding the region spirally as shown in FIG. 8 from the coordinate region. It is also possible to switch or create. At that time, the code sequence for each frame needs to be consistent on the encoding side and the decoding side by inserting information representing the important center coordinates after the frame synchronization signal. Furthermore, by preparing a plurality of the same rearrangement table on the encoding side and the decoding side, and sending them to the receiving side together with the number of the rearrangement table used adaptively according to the image on the encoding side, It is also possible to switch the rearrangement table.

  For example, when a variable-length code capable of bidirectional decoding is used, it is not necessary to divide the code sequence in the forward direction and the reverse direction, and decoding can be performed from the forward direction and the reverse direction as one code sequence. At that time, it is possible to increase the decoding probability of the important part by arranging the code strings as shown in FIGS. 4, 6, 7 and 8, respectively. As an example, FIG. 9 shows how code strings are arranged when a variable-length code capable of bidirectional decoding is used. It is assumed that the blocks with numbers in the drawing represent the blocks corresponding to the macroblocks at the time of encoding.

Second embodiment

  Next, a moving picture decoding apparatus and method according to the second embodiment will be described. FIG. 10 is a block diagram showing an encoding apparatus according to the second embodiment. The difference from FIG. 3 is that, as shown in FIG. 11, the first embodiment divides the entire screen into two area groups, and the code string divider 110 divides the information into corresponding code strings. On the other hand, in this embodiment, instead of each of the first and second code strings including all the areas of the entire screen, only one of the code strings in each area is divided into two by the DCT coefficient string divider 220. Is to include.

  Thereby, even if one information is lost, if the other information can be decoded, an image having a certain level of quality can be decoded in each region even if it is slightly inferior to the normal quality. Compared to the above, uniform image quality can be obtained on the entire screen. In addition to this, the information is not limited to the DCT coefficient, and each other information can be divided into two and encoded.

  Examples of the configuration for dividing the information of each area into two include the following. The coefficients obtained by performing discrete cosine transform on the prediction error signal are conventionally encoded by zigzag scanning as shown in FIG. 33, whereas in the present invention, odd-numbered coefficients and even-numbered coefficients are used as shown in FIG. Divide into coefficients. Further, as shown in FIG. 13, it is possible to divide into a portion that strongly reflects the correlation in the horizontal direction and a portion that strongly reflects the correlation in the vertical direction. Furthermore, while the above example (FIG. 12) divides the DCT coefficient sequence based on a normal zigzag scan, the present invention can also be configured to divide the DCT coefficient based on a scan that matches the image. Is possible.

  FIG. 14 shows a moving picture decoding apparatus corresponding to the moving picture number coding apparatus of FIG. The code sequence transmitted / accumulated from the encoding device detects the frame synchronization signal, and then decodes the prediction error signal for each region based on the information in the encoding order table 1103, and the inverse quantizer 1112 performs inverse quantization and inverse quantization. A series of processes called inverse discrete cosine transform is performed by the DCT unit 1113. At this time, the read code string is stored in the code string memory 1116 at the same time. When the decoding of the front first code string is completed, if any error occurs and decoding becomes impossible, the decoding process is stopped and the code string memory 1116 stores the code until the frame synchronization signal of the next frame. Keep accumulating rows. When the frame synchronization signal of the next frame is found, the code string rearranger 1117 reads the contents of the code string memory 1116 reversely bit by bit from the end, and performs decoding. Decoding is performed in the same manner as the front first code string. When the decoding is completed or cannot be decoded due to an error, the decoding process is terminated, and the front first information and the rear second information can be decoded. The combined information is used as decryption information.

  Then, an adder 1114 adds the prediction signal to obtain a reproduced image. The reproduced image is output to the outside of the apparatus and recorded in the frame memory 1105. By decoding a code string in which the code string is decomposed into two as described above and one is arranged in the forward direction and the other is arranged in the reverse direction, even if there is an error in one and decoding becomes impossible, the other Decoding can be performed without any influence. Further, since the encoding order is changed by rearranging the regions, even when an error is detected by the error detectors 1126 and 1127, a region that cannot be decoded is predicted from the decoded region and interpolated. This is easier than the conventional method.

  Next, a description will be given of an encoding apparatus when transmission is performed using two transmission paths having different error rates. When transmitting on transmission lines with different error rates, important information is transmitted that has a significant effect on a high-quality transmission line with a low error rate, and less important on transmission lines with a high error rate. Hierarchical means such as transmitting information are often used. Therefore, in the same way as in the first and second embodiments, the area of the information transferred to the transmission path with a low error rate (upper layer) and the transmission path with a high error rate (lower layer) is described. By rearranging, dividing the code string, and rearranging the code string, it is possible to cope with hierarchical coding in the same manner. FIG. 15 shows an example of a code string when combined with hierarchization. In this example, important information such as mode information and motion vector is put in the upper layer among the encoded data obtained by rearranging the regions, and the prediction difference signal is put in the lower layer.

  Then, the code string is divided into two in each layer, and one is rearranged in the forward direction from the front to the rear and the other in the reverse direction from the rear to the front. When the code string is divided as in the first embodiment, matching between the upper layer and the lower layer is more correctly performed by performing the same division method in the upper layer and the lower layer. However, when the area of one frame is divided into two, if the code amount is biased to either one, the code string is divided as shown in FIG. 16 considering that the upper layer has a low error rate. It is represented by one code string without being performed, and the code string is divided only in the lower layer. At that time, the number of areas included in (a) and (b) of FIG. 4 is changed so that the code amounts of the two code strings in the lower hierarchy are evenly approached. That is, the number of areas included in FIGS. 4A and 4B is not made equal, but is dynamically changed from mode information or the like of the upper layer. By doing in this way, it can divide | segment so that the code amount of the code sequence divided | segmented into two of the lower hierarchy may become equal.

  In these examples, the prediction error is separated from the other information, but the motion vector is divided into the upper layer and the lower layer according to the code amount of the motion vector, or the code amount of the motion vector is small and the upper layer is low. When there is a margin, it is possible to perform operations such as distributing some of the DCT coefficients of the prediction error signal to the upper and lower layers.

  Further, as shown in FIG. 17, in a system in which highly important information is arranged in the front part for one transmission line, and less important information is subsequently arranged, in FIG. After the code sequences with high importance are arranged in the forward direction, the code sequences with low importance are arranged in the reverse direction, and this is applied so as to be arranged behind the code sequences with high importance (FIG. 17 (c)). Thus, much information can be correctly decoded even when an error occurs compared to the conventional example shown in FIG. At that time, the lower layer information corresponding to the information located at the head of the upper layer code string (forward code string) is arranged so that it is located at the head of the lower layer code string (reverse code string). . In variable-length coding, the probability that the top information is correctly decoded is higher. Therefore, when the method of this embodiment is applied as shown in FIG. 18, the upper part is the front information and the lower part is the rear synchronization signal. The nearer the rear information, the higher the probability of being correctly decoded.

  Therefore, among the methods of encoding for each unit such as a block, in the hierarchical encoding method in which the information for each unit is divided into higher and lower levels and encoded, the lower level information corresponding to the upper head information It is more efficient to place the at the rear (near the rear synchronizing signal). FIG. 19 shows the details. In this case, if not arranged as shown in FIG. 19 (a) but arranged as shown in FIG. 19 (b), there is no upper information corresponding to information correctly decoded in the lower order, so that decoding can be performed at the codeword level. However, the image signal cannot be reproduced from there.

  Next, for applying the moving picture encoding apparatus and decoding apparatus according to the present invention to the hierarchical encoding scheme, the present invention is not applied to each hierarchical code string as in the above embodiment, Each code string of the code string arranged in the forward direction and the code string arranged in the reverse direction may be hierarchized.

  FIG. 20 shows the configuration of a code string for describing an embodiment in which hierarchization is performed for each code string. As shown in FIG. 20, in this embodiment, each of the code strings divided into two is hierarchized. For example, one frame image is interlaced and scanned 2 First, one field is configured as a forward code string, and the other is configured as a reverse code string. Next, in each code string, information necessary for decoding is arranged as upper information, and a residual part is arranged as lower information, for example, and hierarchized.

  Even if one of the hierarchized code strings configured as described above cannot be decoded due to an error or the like, the other can be completely decoded. Therefore, it is possible to implement hierarchical encoding more easily and efficiently than hierarchically classifying the upper information and the lower information. Moreover, if the upper information is not completely decoded, efficient code hierarchization can be realized even in a system that cannot be used even if the lower information is decoded. For example, in the example of FIG. 15, a system in which a synchronization signal is not inserted between the upper information and the lower information, or the upper information There may be a case where pointer information indicating the end of the file is not added. Therefore, in the embodiment of FIG. 15, by inserting a synchronization signal between the upper information and lower information or adding pointer information or the like, unless the lower information is decoded even if the higher information is decoded, Although the problem of not being able to use any information can be improved, the above problem can be solved extremely easily by hierarchical encoding as in this embodiment.

  Next, an embodiment in which the moving image is divided into three code strings and encoded in the moving image encoding device according to the present invention will be described with reference to FIG. In this embodiment, a moving image is encoded based on the above-described concept of hierarchical encoding. For example, important information is divided into two and a first code sequence and a second code sequence are divided. And a third code string is further configured for information not included in the code string of important information. After the first to third code strings are configured by the arrangement according to such importance, specifically, the code string is configured by an arrangement as illustrated in the lower side of FIG. By arranging the codes in such an arrangement, it is possible to enhance protection of important information.

  For example, when the hierarchical coding of the embodiment shown in FIG. 21 is applied to coding of image information, mode information and motion vector information are allocated to the important information section, and these pieces of information are arranged at any position on the screen. It is divided into two code strings in consideration of whether or not In this division, even if one of the two code strings for important information is lost due to an error or the like, the macroblock is selected in a checkered pattern so that the other can compensate for the loss, By making the black block correspond to the first code string and the white block correspond to the second code string, processing such as flashing and concealment is easy. It is also possible to do so.

  In this way, the important information is divided and arranged in the first and second code strings, and a residual part may be assigned to the third code string, but the error tolerance of the third code string itself is improved. Therefore, a fixed length code or a variable length code that can be decoded bidirectionally may be arranged in the third code string.

  The above configuration can be combined with the duplication of the code string. The second code string is the same as or similar to the first code string, and the first code string is normally forward-oriented by configuring the second code string in the arrangement shown in the lower part of FIG. When only an error occurs in the first code string, the second code string is decoded in the reverse direction to obtain complete image information without error. be able to.

  Further, regarding the order of arrangement of the respective code strings and the positional relationship of the screen, for example, the first code string is applied to the corresponding pixel information from the upper left to the lower right of the screen, and the second code string is set to the lower right of the screen. By applying from the upper left to the upper left, even if an error occurs in any of the codewords, the first code string decoded from the upper left to the lower right and the second code string decoded from the lower right to the upper left overlap. Can be used. Thus, for example, even if an error occurs in any part of any code string, it is possible to compensate for each other by decoding the first and second code strings in the forward and backward directions. The probability of information that can be decoded can be improved. In this way, even if both of the sequences in which the pair of code strings are constructed cannot be completely decoded due to errors, they can be complemented to each other, so that decoding can be correctly performed when an error occurs in any of the code words. The ratio of information can be increased.

  In addition, since it is not necessary to newly add a synchronization signal or the like to the head of the duplexed information, the efficiency is improved as compared with the case of simply duplexing. This configuration is not limited to the arrangement shown in FIG. 21, and one of the code strings is a normal code string and the other code string is a spare code string. It is applicable to.

  Next, as an application example of the above-described embodiment, the order of the codewords is not set to the forward direction and the reverse direction in units of code strings, but the order of the codewords is reversed to the forward direction in units of smaller information. You may comprise so that it may become a direction. In any of the above-described embodiments, the code string is divided into two information units according to some standard, one is arranged in the forward direction, and the other is arranged in the reverse direction. However, since the code strings arranged in the reverse direction at the time of encoding need to be output after being stored once in the buffer, the reordering process is performed only after the encoding is completed. There is a problem that the processing operation is delayed and the processing operation is delayed.

  Therefore, instead of encoding with a relatively large code string unit as a unit of information, for example, encoding in the reverse direction with a small unit such as a macroblock unit reduces the delay of the processing operation. be able to. The configuration in such an embodiment is shown in FIG. In FIG. 22, when the information of the macroblocks MB1 to MB10 is divided in half, the macroblocks MB6 to MB10 arranged in the reverse direction are not rearranged together, but are rearranged for each macroblock. It means to become.

  However, in such an embodiment, it is necessary to decode sequentially from the macroblock MB10 using a decoder. However, the normal decoder encodes the difference between the previous macroblock and the subsequent macroblock, and uses this differential code to decode the subsequent macroblock. Since this block is used for decoding the block, it must be decoded in order from the macroblock MB1.

  Therefore, in this embodiment, the additional information is added so that it can be decoded from the subsequent information, thereby overcoming the problem of having to decode in the forward direction as described above. For example, the actual values of the motion vector and the quantization width of the macroblock MB10 so that the information that encodes the difference from the previous information such as the motion vector and the quantization width can be decoded from behind. As additional information, decoding can be started even from the macroblock MB10. After that, since the information about the difference from the macroblock MB9 is included in the information of the macroblock MB10, the macroblock MB9 is decoded using this information. Similarly, a group of information can be completely decoded by sequentially decoding from the reverse direction like macroblocks MB8, MB7, and MB6. This additional information does not necessarily have to be added to the position shown in FIG. 22, and any position can be used as long as the position can be specified before decoding the codewords arranged in the reverse direction. It is possible to add.

  As mentioned above, the embodiment of the hierarchical coding has been described. However, the method described here is not particularly intended to specify the arrangement of the code strings when coding the information having a difference in importance such as the upper hierarchy and the lower hierarchy. Of course, the same method can be applied to two code strings having the same importance.

  In addition, the coding apparatus according to the present invention is not only applied to the upper and lower two layers, but the present invention can also be applied to a collection of multi-layer information as described above.

  In the above embodiment, the code strings are rearranged in units of frames and synchronized with the frame synchronization signal. However, there is no need for the frame synchronization signal, and the first code string before the first code string is used. If there is a synchronization signal after the code string of 2, it can be used in all cases. For example, when a plurality of synchronization signals exist in one frame (such as when a synchronization signal is inserted in each row of FIG. 30A), the conventional method is as shown in FIG. As shown in FIG. 30 (c), the inter-frame sync signal inserted every time is rearranged in the region from the sync signal to the sync signal as a unit, and the code string division and the code string rearrangement are performed within one line. Is also possible. That is, the conventional rearrangement as in the conventional example of FIG. 30B can be performed as shown in FIG.

  FIG. 23 is a diagram in which a method of arranging one frame from important regions is applied to a method of configuring one frame by a plurality of sync blocks. As shown in FIG. 34, as in this method, a synchronization signal is inserted in a frame other than the frame synchronization signal (PSC) to reduce the amount of information that cannot be used even in the case of an error. ing. As shown in FIG. 23, for example, each sync block obtained by dividing the frame by the sync signal is encoded in the order as shown in FIG. 23 (a) so as to correspond to the first sync block, and in the order (b). By making the encoded one correspond to the second sync block, even if one of them is lost due to an error, the other can compensate. Further, by arranging from the important blocks, the probability that the important blocks can be correctly decoded increases.

  In the above embodiment, bi-directional decoding is possible by attaching a synchronization code to both ends of the code string. However, the present invention can identify the last part of the code string before decoding the code string. If present, the synchronization signal is not particularly necessary. Hereinafter, an embodiment according to a method that does not use the synchronization code will be described.

  In the case of an encoding system in which information of a predetermined unit such as information for one frame or several macroblocks is output after being fixed-length encoded, the decoding side does not rely on a synchronization signal or the like in the forward direction. The heads of the arranged code strings and the code strings arranged in the reverse direction can be specified. FIG. 24 shows an example in which the encoding apparatus according to the present invention is applied to this fixed-length encoded code string. In this example, an embodiment is shown in which the present invention is applied to a code string that is fixed in length so that N macroblocks (MB) are combined into m bits. In this embodiment, as shown in FIG. 25, since the head of the code sequence arranged in the reverse direction always exists every m bits, it is necessary to interpolate the synchronization signal in the code sequence as described above. Is eliminated, and can be correctly decoded regularly.

  Next, unlike the above configuration, even in a system in which information is not a fixed length in the predetermined unit, it is possible to specify the head position of a code string arranged in the reverse direction without using a synchronization signal. Embodiments will be described. The system according to this embodiment is configured to output a code string by adding information that can identify the position where the head of the code string in the reverse direction is located. FIG. 26 shows a code string according to this embodiment. By placing pointer information indicating the head at a frame header that can be synchronized or at a position equivalent to this, the decoder uses this pointer information to identify the head of the code sequence arranged in the reverse direction. Then, these code strings are decoded. As described above, by configuring so that the end of the code string can be specified in a predetermined unit instead of the synchronization signal, the present invention can be implemented even with a variable-length coding system.

  Next, the high-speed playback operation in the decoding apparatus in the first embodiment and the second embodiment will be described. In the case of high-speed forward reproduction, after receiving a frame synchronization signal indicating the head of the frame, only the first code string in the front described in the forward direction is decoded by the decoding process and the process proceeds to the next frame. Information corresponding to the second code string rearranged in the reverse direction corresponds by performing interpolation or the like using information decoded by the first code string.

  In the case of high-speed reverse reproduction, as in the case of high-speed forward reproduction, the frame synchronization signal is searched for in the reverse direction, and only the second code string that has been rearranged in the reverse direction after receiving the frame synchronization signal is decoded. Do. At this time, when motion compensation adaptive predictive coding is used, an image of the previous frame is necessary, and therefore it is necessary to select only a frame that does not use motion compensation.

  FIG. 27 shows a third embodiment in which the moving picture coding and decoding apparatus according to the present invention is applied to a wireless communication system. In FIG. 27, the wireless communication system includes an image transmission system 20 and an image reproduction system 30, and image transmission and reception are performed via a base station 41 provided with a network 40.

  The image transmission system 20 includes an image signal input unit 21, an information source encoding unit 22 including an error resilience processing unit 23, a transmission path encoding unit 24, and a radio unit 25. The unit 22 performs discrete cosine transform (DCT), quantization, and the like, and the traditional path encoding unit 24 performs error detection and correction of encoded data.

  The image reproduction system 30 includes a radio unit 31, a traditional path decoding unit 32, an information source decoding unit 33 including an error resilience processing unit 34, and an image signal output unit 35.

  FIG. 28 shows an example to which the moving picture encoding apparatus and decoding apparatus according to the present invention are applied. As shown in the figure, the laptop is connected via the base stations 41, 42, 43 of the wireless communication network 40. A moving image is transmitted and received by a terminal 50 such as a personal computer 51 or a desktop personal computer 52.

The block diagram which shows the basic concept of the moving image encoder by this invention. The figure which shows the encoding sequence formed by the encoding apparatus of FIG. 1 is a block diagram of a video encoding / decoding device according to a first embodiment of the present invention. The figure which shows an example of the area | region rearrangement order of the moving image encoding / decoding apparatus of this invention. The figure which shows the code sequence encoded by the moving image encoding / decoding apparatus of this invention. The figure which shows the example of the area | region rearrangement order of the moving image encoding / decoding apparatus of this invention. The figure which shows the example of the area | region rearrangement order of the moving image encoding / decoding apparatus of this invention. The figure which shows the example of the area | region rearrangement order when the center area | region of this invention is designated. The figure which shows how to arrange | position the code sequence at the time of making this invention respond | correspond to a bi-directionally decodable code. The block diagram of the moving image encoding / decoding apparatus by 2nd Embodiment of this invention. The figure which shows the difference between 1st Embodiment of this invention and 2nd Embodiment. The figure which shows the example of the scan order of the DCT coefficient of this invention. The figure which shows the example which divides | segments the scan of the DCT coefficient of this invention into the order which considered the correlation of the horizontal and vertical direction. The block diagram of the moving image decoding apparatus by embodiment of this invention. The figure which shows the Example corresponding to the hierarchical coding of this invention. The figure which shows the Example corresponding to the hierarchical coding of this invention. The figure which shows the example of the hierarchical coding system of this invention, and the conventional hierarchical coding system. The figure which shows embodiment corresponding to the hierarchical coding of this invention. The figure which shows the time of error generation of embodiment corresponding to the hierarchical coding of this invention. The figure which shows embodiment which applied this invention to hierarchical encoding. The figure which shows the structure of the code sequence at the time of making this invention respond | correspond to three code sequences. The figure which shows the structure of the code sequence when the delay of a processing operation can be decreased by this invention. The figure which applied the system arranged from the important area of the present invention to the system which constitutes one frame by a plurality of sync blocks. The figure which shows the structure of the code sequence in embodiment which applied this invention to the fixed-length code. The figure which shows the structure of the other code sequence of embodiment which applied this invention to the fixed-length code. The figure which shows the code sequence of embodiment implemented using additional information. The figure which shows embodiment which applied the moving image encoding apparatus and decoding apparatus which concern on this invention to a radio | wireless communications system. The figure which shows embodiment which applied the moving image encoding / decoding apparatus which concerns on this invention to the radio | wireless communications system. The block diagram of the conventional moving image encoder. The figure which shows the encoding order of the conventional moving image encoder. The figure which shows the code sequence encoded by the conventional moving image encoder. The block diagram of the conventional moving image decoding apparatus. The figure which shows the example of the scanning order of the conventional DCT coefficient. The figure which shows the example which inserted the synchronizing signal in the conventional frame.

Explanation of symbols

101, 102, 103 Area dividers 102, 202 Area rearrangers 103, 203, 1103 Area rearrangement tables 104, 204, 304, 1104, 1204 Motion compensators 105, 205, 305, 1105, 1205 Frame memories 106, 206 , 306 Difference units 107, 207, 307 Discrete cosine transformers 108, 208, 308 Quantizers 109, 209, 309 Multiplexer 110 Code stream divider 111, 211 Code stream rearranger 112, 212, 312, 1112 1212 Inverse quantizer 113, 213, 313, 1113, 1213 Inverse discrete cosine transformer 114, 214, 314, 1114, 1214 Adder 1115, 1215 Demultiplexer 1116 Code string memory 1117 Code string rearranger 1119 Image signal Sorter 220 DCT coefficient sequence divider 221 DCT coefficient sequence synthesizers 122, 123, 222, 223, 322, 323 Variable length encoders 1124, 1125, 1224, 1225 Variable length decoders 1126, 1127 Error detectors 131, 231, 331 Input image signal 133,233,333 Prediction signal 134,234,334 Prediction residual signal

Claims (10)

  Decoding means for decoding the image information of each area divided into a plurality of areas for each frame and decoding the area rearrangement table describing the information indicating the rearrangement order of each area, and the decoding A moving picture decoding apparatus comprising: a region rearranging unit that rearranges the image signals of the respective regions decoded according to the region rearrangement order described in the region rearrangement table decoded by the unit into a correct rearrangement order.   The moving image decoding apparatus according to claim 1, wherein the region rearranging unit rearranges the image signals of each region obtained by decoding the image signal input for each frame in units of a plurality of macroblocks.   The moving picture decoding apparatus according to claim 1, wherein the information indicating the rearrangement order described in the area rearrangement table area is an order number indicating the area rearrangement order.   The moving image according to any one of claims 1 to 3, wherein the area rearranging means rearranges the area with a screen area where important information is located at the head of the rearrangement order. Decoding device.   The decoding means decodes the first code string arranged in the forward direction from the front to the rear in the image information of each area encoded from each divided area in the forward direction from the front to the rear. 5. The moving picture decoding apparatus according to claim 1, wherein the second code string arranged in the forward direction in the reverse direction is decoded in the reverse direction from the rear to the front.   A decoding step for decoding image information of each region divided into a plurality of regions for each frame and decoding a region rearrangement table describing information indicating a rearrangement order in units of regions; And a region rearrangement step of rearranging the image signals of the respective regions decoded according to the region rearrangement order described in the region rearrangement table in a correct rearrangement order.   7. The moving picture decoding method according to claim 6, wherein the first decoding step decodes an image signal input for each frame in units of a plurality of macroblocks.   8. The moving picture decoding method according to claim 6, wherein the information indicating the rearrangement order described in the area rearrangement table area is an order number indicating the area rearrangement order.   The moving image according to any one of claims 6 to 8, wherein in the region rearrangement step, the region is rearranged with a screen region where important information is located at the head of the rearrangement order. Decryption method.   In the decoding step, the first code string arranged from the front to the rear in the image information of each area encoded from each divided area is decoded in the forward direction from the front to the rear, and from the rear to the front. 10. The moving picture decoding method according to claim 6, wherein the second code strings arranged in the reverse direction are decoded in the reverse direction from the rear to the front.

Images