JP2007235974A - Image editing apparatus and image editing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To perform editing at high speed when constituting, by means of editing, a bitstream C at least partially connecting bitstreams A and B resulting from MPEG coding, respectively. <P>SOLUTION: A data accumulation amount D<SB>A</SB>or D<SB>B</SB>of a VBV buffer is calculated, respectively at an ending point A<SB>out</SB>of a scene A' in a part of the bitstream A or at a starting point B<SB>in</SB>of a scene B' in a part of the bitstream B and based on a difference between the data accumulation amounts, a data amount of a portion corresponding to a connecting point of the bitstreams A' and B' in the bitstream C is adjusted. Namely, in D<SB>A</SB>>D<SB>B</SB>, a stuffing code is added to a picture at the ending point A<SB>out</SB>. In D<SB>A</SB><D<SB>B</SB>, on the other hand, a skipped P-picture is inserted just behind a picture at the ending point A<SB>out</SB>. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image editing apparatus and an image editing method, and more particularly to an image editing apparatus suitable for use in, for example, a so-called cut editing of a bitstream obtained by encoding an image with MPEG (Moving Picture Experts Group). The present invention relates to an image editing method.

  FIG. 11 shows a configuration of an example of a conventional image editing apparatus.

  The data storage device 21 is composed of, for example, a hard disk or the like. For example, bitstreams A and B of image data compression-encoded in accordance with ISO-11172-2 (so-called MPEG1) standard are used as editing target materials. It is recorded. In addition, the data storage device 21 includes a bitstream C obtained by editing the bitstreams A and B under the instruction of the controller 124 and connecting at least a part of the bitstreams A and B. Are also being recorded.

  The bit counter 22 counts the data amount (bit amount) of the bit stream read from the data storage device 21 and supplies it to the controller 124 at a timing instructed by the controller 124. The bit counter 22 is also configured to output the bit stream from the data storage device 21 to the decoder 23 as it is. The decoder 23 decodes various parameters included in the bit stream supplied via the bit counter 22 under the control of the controller 124 and outputs the decoded parameters to the controller 124. That is, the decoder 23 decodes, for example, a sequence header (Sequence Header), a GOP (Group Of Picture) header (GOP Header), and a picture header (Picture Header) included in the bit stream, and is arranged in each header. The information (parameter) is output to the controller 124.

  Based on the outputs of the bit counter 22 and the decoder 23, the controller 124 configures the bit stream C from the bit streams A and B recorded in the data storage device 21 using the memory 25 as necessary, and stores the data ( Recording) Recording is performed on the device 21.

  Next, the operation will be described.

  Here, for example, as shown in FIG. 12, a part of the scene A ′ is cut out from the bit stream A, and a part of the scene B ′ is cut out from the bit stream B, and the scene B ′ is cut after the scene A ′. It is assumed that the bit stream C is configured by connection.

Further, the start point or end point of the scene A ′ is described as A _in or A _out , respectively, and the start point or end point of the scene B ′ is described as B _in or B _out , respectively.

Further, since MPEG1 stipulates that editing of a bit stream of an image must be performed with GOP as one unit, start points A _in and B _in are the first picture of GOP, end points A _out and B _out is to point to the last picture of the GOP, it is assumed that preset.

Therefore, it is necessary to detect the GOP in order to detect the start points A _in and B _in and the end points A _out and B _out , but the GOP is detected by, for example, the GSC ( Group Start Code). However, when the end point A _out or B _out is set at the end of the bit stream A or B, the next GOP does not exist, but here it is assumed that the process is performed.

  In the controller 124, as shown in the flowchart of FIG. 13, first, in step S101, the sequence headers of the bit streams A and B are read. That is, the controller 124 controls the data storage device 21 to supply the bit stream A to the decoder 23 via the bit counter 22 and decode the sequence header therein. Similarly, the controller 124 also decodes the sequence header of the bit stream B.

  The sequence headers of the decoded bit streams A and B are supplied to the memory 25 via the controller 124 and stored.

  Thereafter, in step S102, the controller 124 compares the sequence headers for the bit streams A and B stored in the memory 25, thereby cut-editing the two bit streams A and B into a bit stream C. Determine whether it is possible. That is, since the parameters before the LIQM (Load Intra Quantize Matrix) among the parameters arranged in the sequence headers for the bitstreams A and B cannot be edited, they cannot be edited. Then, it is determined whether or not these parameters match.

  If it is determined in step S102 that the bitstreams A and B cannot be edited, that is, the parameters arranged in the sequence header for the bitstreams A and B do not match before the LIQM. In this case, the controller 124 displays on the monitor (not shown) that editing cannot be performed and ends the process.

  Further, when it is determined in step S102 that the bitstreams A and B can be edited, that is, the parameters arranged in the sequence header for the bitstreams A and B match those before the LIQM. If YES in step S103, the controller 124 reads the bit stream A from the data storage device 21 in units of pictures, for example, and supplies it to the bit counter 22 and further to the decoder 23.

  In the bit counter 22, the bit amount of the bit stream A is counted, and in the decoder 23, when the supplied bit stream A includes a sequence header, it is decoded and supplied to the controller 124.

  The controller 124 monitors the output of the decoder 23 and determines whether or not a sequence header is output from the decoder 23 in step S104. If it is determined in step S104 that the sequence header has been output, the process proceeds to step S105, where the sequence header for the bit stream A stored in the memory 25 is updated to the latest sequence header output by the decoder 23, The process proceeds to S106.

Further, in step S104, if the sequence header is determined not to be output, by skipping step S105, the process proceeds to step S106, the controller 124, from the data storage device 21, read out a picture of the start point A _in It is determined whether it has been done. In step S106, the picture at the start point A _in is, if it is determined not yet read, the flow returns to step S103, the controller 124 continues to read the bit stream A.

Further, in step S106, if the picture at the start point A _in is determined to have been read, the process proceeds to step S107, the controller 124, the latest sequence header of the bit stream A which is stored in the memory 25, bit The first sequence header for stream C is recorded in the data storage device 21, and the process proceeds to step S108. In step S108, the controller 124 reads the GOP header of the GOP to which the picture of the starting point A _in belongs from the data storage device 21, causes the decoder 23 to decode it, and receives the decoding result.

  Then, the controller 124 proceeds to step S109, and determines whether or not CG (Closed GOP) arranged in the GOP header received from the decoder 23 is zero. If it is determined in step S109 that CG is 0, that is, if the GOP is a so-called open GOP, the process proceeds to step S110, and BL (Broken Link) arranged in the GOP header is set to 1, Proceed to step S112.

That is, when the GOP including the picture of the starting point A _in is an open GOP, the link with the previous GOP is cut by editing, so the first B picture in the display order in the GOP It cannot be decoded correctly. Therefore, BL indicating that is set to 1.

  On the other hand, if it is determined in step S109 that CG is not 0, that is, if the GOP is a so-called closed GOP, the process proceeds to step S111, and the BL arranged in the GOP header is set to 0, and the process proceeds to step S112. move on.

  That is, when the GOP is a closed GOP, the first picture in the display order in the GOP is an I picture or a B picture that is not forward-predicted, so the preceding GOP is not used for decoding. Therefore, BL is set to 0.

  In step S112, the controller 124 records the GOP header in which BL is set in step S110 or S111 as the GOP header for the bitstream C following the sequence header recorded in step S107, and the process proceeds to step S113.

In step S113, the reading of the bit stream A by the controller 124 is continued, and the process proceeds to step S114, where it is determined whether or not the picture of the end point A _out has been read from the data storage device 21. If it is determined in step S114 that the picture of the end point A _out has not been read yet, the process proceeds to step S115, and the controller 124 reads the data read in step S113 as data for the bitstream C. Is recorded in the data storage device 21.

  In step S116, the controller 124 determines whether or not the data read in step S113 is a sequence header by referring to the output of the decoder 23. If it is determined in step S116 that the data read in step S113 is a sequence header, the process proceeds to step S117, and the sequence header for the bit stream A stored in the memory 25 is the latest output from the decoder 23. It is updated to the sequence header (sequence header read in step S113). Then, the process returns to step S113, and the reading of the bit stream A is continued.

  If it is determined in step S116 that the data read in step S113 is not a sequence header, step S117 is skipped, the process returns to step S113, and the reading of bitstream A is continued.

On the other hand, if it is determined in step S114 that the picture of the end point A _out has been read, the process proceeds to step S121 in FIG. 14 and the data read in step S113 is stored as data for the bit stream C. The data is recorded in the device 21 and the process proceeds to step S122. In step S122, the reading of the bit stream A is continued, and the process proceeds to step S123. The controller 124 determines whether the data read in step S122 is a sequence header, a GOP header, or an SEC (Sequence End of Code). It is determined by referring to the output of the decoder 23.

  If it is determined in step S123 that the data read in step S122 is not one of the sequence header, GOP header, or SEC, the process returns to step S121 and the above-described processing is repeated.

  If it is determined in step S123 that the data read in step S122 is one of the sequence header, GOP header, or SEC, the controller 124 does not record the data. In step S124, reading of the bit stream B from the data storage device 21 in units of pictures, for example, is started.

  In step S125, the controller 124 refers to the output of the decoder 23 to determine whether the data read in step S124 is a sequence header. If the controller 124 determines that the data is a sequence header, step S126 is performed. The sequence header for the bit stream B stored in the memory 25 is updated to the latest sequence header output by the decoder 23, and the process proceeds to step S127.

Further, in step S125, the case where the data read in step S124 is determined not to be a sequence header, skips step S126, the process proceeds to step S127, the controller 124, from the data storage device 21, the start point B _in It is determined whether the current picture has been read out. In step S127, the picture of the start point B _in the, if it is determined not yet read, the flow returns to step S124, the controller 124 continues to read the bit stream B.

Further, in step S127, if the picture at the start point B _in is determined to have been read, the process proceeds to step S128, the controller 124, the most recent sequence header each other of the bit streams A and B are stored in the memory 25 Are compared to determine whether the LIQM does not match, and whether the LNIQM (Load Non Intra Quantizer Matrix) does not match.

  If it is determined in step S128 that the LIQMs arranged in the latest sequence headers for the bitstreams A and B do not match or the LNIQMs do not match, the process proceeds to step S129, and the controller 124 stores in the memory 25. The latest sequence header for the stored bit stream B is recorded in the data recording device 21 as the sequence header of the bit stream C, and the process proceeds to step S130.

If it is determined in step S128 that the LIQMs arranged in the latest sequence headers for the bitstreams A and B and the LNIQMs match each other, the process skips step S129 and proceeds to step S130. , the controller 124, the picture at the start point B _in read in step S124, the corresponding data storage amount of VBV (Video Buffering Verifier) buffer to assume the reception side (decoder side) is defined in the MPEG VD Calculate (VBV Delay).

  Here, VD will be described later.

  In step S131, the controller 124 determines whether the VD value calculated in step S130 is inappropriate. If it is determined in step S131 that the value of VD is inappropriate, the bitstreams A and B can only be edited after being MPEG-decoded, so the processing ends.

Further, in step S131, the case where the value of VD is not determined to be inappropriate, the flow proceeds to step S132, the picture header of the picture of the read start point B _in at step S124, the is VD calculated in step S130 Overwriting is performed and the process proceeds to step S133. In step S133, (in this case, a picture of the read start point B _in Step S124) picture VD of the picture header has been rewritten as data of the bit stream C, is recorded in the data storage device 21.

Thereafter, the controller 124 continues to read the bit stream B from the data storage device 21 in step S134, proceeds to step S135, and determines whether the picture at the end point B _out has been read from the data storage device 21. Is done. In step S135, the picture of the end point B _out is, if it is determined not yet read, the flow returns to step S130, processing similar to the case described above is performed.

  That is, in step S130, VD (VBV Delay) corresponding to the amount of data stored in the VBV buffer is calculated for the picture read out in step S134, and the process proceeds to step S131 to determine whether or not the value of VD is inappropriate. Is done. If it is determined in step S131 that the value of VD is inappropriate, the process is terminated. If it is determined that the value of VD is not appropriate, the process proceeds to step S132, and is read in step S134. The VD calculated in step S130 is overwritten on the picture header of the picture, and the process proceeds to step S133. In step S133, the picture in which the VD of the picture header is rewritten (here, the picture read in step S134) is recorded in the data storage device 21 as data about the bit stream C.

On the other hand, in step S135, if the picture at the end point B _out is determined to have been read, the process proceeds to step S136, the picture (in this case, a picture at the end point B _out) is, as the data of the bit stream C, It is recorded in the data storage device 21. In the process of step S136, the same process as in steps S130 to S132 (VD is calculated, it is determined whether it is inappropriate, and if not appropriate, the VD of the picture header is updated. Processing) is performed.

  Thereafter, the process proceeds to step S137, where SEC is recorded as the last data for the bit stream C, and the process ends.

  By the way, MPEG stipulates that the bit stream must be configured so that the VBV buffer does not overflow or underflow.

  Here, FIG. 15 shows an example of the transition of the data accumulation amount of the VBV buffer. In the figure, the horizontal axis or the vertical axis indicates time or data accumulation amount, respectively.

  A bit stream obtained as a result of MPEG encoding is accumulated in the VBV buffer at a predetermined constant bit rate. A value obtained by rounding up a value of 400 bps (Bit Per Seconds) or less of the constant bit rate and dividing by 400 is arranged as BR (Bit Rate) in the sequence header of the bit stream.

  Reading of the bit stream from the VBV buffer (that is, input to the decoder) is performed according to PR (Picture Rate) arranged in the sequence header. That is, when the decoding time that comes in a certain period according to PR is reached, data about the picture to be decoded at that decoding time is instantaneously read from the VBV buffer and supplied to the decoder.

Therefore, the data amount corresponding to the bit rate / picture rate is accumulated in the VBV buffer between the time when a certain picture is read and the time when the next picture is read. That is, for example, the amount of data shown in FIG. 15 Oite D ₁ is represented by a bit rate / picture rate.

Further, at a certain decoding time, SHC (Sequence Header Code) indicating the start of the sequence header (000001B3 in hexadecimal in MPEG1), GSC (Group Start Code) indicating the start of the GOP header (000001B8 in hexadecimal in MPEG1). ), Or the first bit of any one of PSC (Picture Start Code) (MPEG1 in hexadecimal number 00000100) representing the start of the picture header to the immediately preceding bit of the next SHC, GSC, or PSC Are immediately read from the VBV buffer. In FIG. 15, the portion indicated by D ₂ shows how data is read from the first bit of SHC.

  The amount of data stored in the VBV buffer can be recognized from the VD arranged in the picture header described above. That is, the VD stores the data at the bit rate corresponding to BR and assumes that the data storage amount is the original state when it is assumed that the last bit of the PSC has been read from the VBV buffer. Is counted with a 90 kHz clock.

When editing as described above, the amount of data stored in the VBV buffer at the end point A _out of the bit stream A and the amount of data stored in the VBV buffer at the start point B _in of the bit stream B are usually For example, there is a gap as shown in FIG. Therefore, the order to correct the gap, the VD for a picture at the start point B _in, there is a need to be changed to the amount of accumulated data is aligned at the start point B _in the end point A _out.

Specifically, the VD for the picture at the start point B _in needs to be VD ′ (B _in ) expressed by the following equation.

VD ′ (B _in ) = VD (A _out )
+ (HL (B _in ) −HL (A _out ))
× 90000 / br
... (1)
However, in Expression (1), VD (A _out ) represents VD at the end point A _out , and HL (B _in ) or HL (A _out ) represents the picture at the start point B _in or the end point A _out . This represents the amount of bits from the first bit of the sequence header or GOP header to the last bit of the PSC. Br represents a bit rate obtained from BR.

Here, in step S130 of FIG. 14, VD (VD ′ (B _in )) for the picture at the start point B _in is calculated according to the equation (1).

When the VD for the picture at the start point B _in is changed to VD ′ (B _in ) obtained by the equation (1), the VD included in the subsequent bit stream B needs to be changed. This is obtained by adding VD ′ (B _in ) −VD (B _in ) to the original VD, and changing this value.

Here, in step S130 of FIG. 14, VDs other than the picture at the start point B _in are obtained in this way.

When the capacity of the VBV buffer is expressed as S _VBV, it can be said that the VD after the change becomes a value of 0 or less or exceeds 90,000 × S _VBV / br or a hexadecimal value exceeding FFFE. Is not allowed because it will underflow or overflow. Here, in step S131 of FIG. 14, it is determined whether or not the VD calculated in step S130 has such a value.

Incidentally, in the case of editing as described above, basically, it is necessary to rewrite all the starting point B _in the subsequent VD, it was not interfere with high-speed processing.

  Furthermore, when the VD has the above value, editing cannot be performed unless the bitstream is decoded, which is troublesome.

  In calculating VD, a bit rate br is required. This bit rate br is obtained from BR arranged in the sequence header. However, as described above, since BR is a rounded value of the true bit rate in 400 bps units, VD may not be obtained with high accuracy.

  The present invention has been made in view of such circumstances, and is intended to enable high-speed editing of a compression-encoded bitstream in that state.

  An image editing apparatus according to one aspect of the present invention edits at least a part of first and second bitstreams obtained by performing compression coding on at least an image, and each of the first and second bitstreams. Is an image editing apparatus that constitutes a third bit stream connected to the data, and the compression encoding is performed in consideration of the data accumulation amount of the virtual buffer assumed on the receiving side, and the accumulation amount relating to the data accumulation amount In the case where information is included in the first and second bitstreams, the first connection point of the first bitstream that connects with the second bitstream, and the second bitstream, A data storage amount of the virtual buffer at a second connection point connected to the first bit stream is calculated based on the storage amount information. Based on the difference between the data accumulation amounts at the first and second connection points calculated by the data accumulation amount calculation means and the data accumulation amount calculation means. Data amount adjusting means for adjusting the data amount of the portion corresponding to the connection point of the second bit stream.

  The first and second bitstreams are obtained by compression-coding at least images according to MPEG (Moving Picture Experts Group) standards, and the accumulated amount information is data in a VBV (Video Buffering Verifier) buffer. It is VD (VBV Delay) included in the picture header, which corresponds to the accumulation amount.

  In the data amount adjusting means, when the data accumulation amount at the first connection point is larger than the data accumulation amount at the second connection point, a stuffing code is added to the picture data at the first connection point. Can be inserted.

  The stuffing code can be inserted into the data amount adjusting means so that the data accumulation amounts at the first and second connection points are equal.

  When the data accumulation amount at the first connection point is smaller than the data accumulation amount at the second connection point, the data amount adjusting means includes a skipped P picture (after the picture data at the first connection point). Data about Skipped P-picture) can be inserted.

  The data amount adjusting means can insert the skipped P picture until the data accumulation amount after inserting the skipped P picture becomes equal to or greater than the data accumulation amount at the second connection point.

  When the skipped P picture is inserted into the data amount adjusting means and the data storage amount after the insertion becomes larger than the data storage amount at the second connection point, the data storage amounts are equal. As can be seen, a stuffing code can be inserted.

  The image editing apparatus further includes bit rate estimating means for estimating the bit rates of the first and second bit streams, and the data accumulation amount calculating means includes the bit estimated by the bit rate estimating means. Based on the rate and the storage amount information VD, the data storage amount of the VBV buffer can be calculated.

  According to an image editing method of one aspect of the present invention, at least a part of each of the first and second bitstreams is edited by editing first and second bitstreams obtained by performing compression coding on at least an image. Is a video editing method for configuring a third bitstream connected to each other, wherein the compression encoding is performed in consideration of the data accumulation amount of the virtual buffer assumed on the receiving side, and the accumulation amount relating to the data accumulation amount In the case where information is included in the first and second bitstreams, the first connection point of the first bitstream that connects with the second bitstream, and the second bitstream, A data accumulation amount of the virtual buffer at a second connection point connected to the first bit stream is calculated based on the accumulation amount information, Based on the difference between the data accumulation amounts at the first and second connection points, the data amount of the portion corresponding to the connection point of the first and second bit streams in the third bit stream is adjusted. Includes steps.

  In one aspect of the present invention, a first connection point of the first bit stream connected to the second bit stream, and a second connection point of the second bit stream connected to the first bit stream. The data accumulation amount of the virtual buffer at is calculated based on the accumulation amount information, and based on the difference between the data accumulation amounts at the first and second connection points, the first and second in the third bitstream The amount of data corresponding to the connection point of the bitstream is adjusted.

  As described above, according to one aspect of the present invention, a compression-encoded bitstream can be edited. In particular, for example, a compression-encoded bitstream can be edited at a high speed in that state. Can do.

  FIG. 1 shows a configuration example of an embodiment of a nonlinear editing apparatus to which the present invention is applied.

  This non-linear editing apparatus is configured based on a computer, for example. That is, the microprocessor 1 executes an application program recorded on the hard disk 12 under the control of the operating system recorded on the hard disk 12, thereby reproducing the edited image and sound and reproducing the edited image and sound. A predetermined process such as a process is performed. The main memory 2 stores a program executed by the microprocessor 1 and data necessary for the operation of the microprocessor 1. The frame buffer 3 is composed of, for example, a DRAM (Dynamic Random Access Memory) and stores an image generated by the microprocessor 1. The bus bridge 4 controls data exchange between an internal bus and an expansion bus such as a PCI (Peripheral Component Interconnect) local bus.

  The above microprocessor 1, main memory 2, frame buffer 3, and bus bridge 4 are connected to each other via an internal bus, and the remaining blocks are connected to each other via an expansion bus. The bus bridge 4 is connected to both the internal bus and the expansion bus.

  The tuner 5 receives a television broadcast signal broadcast using, for example, a terrestrial wave, a satellite line, or a CATV network. Here, an image, a sound, or the like received by the tuner 5 can be an object to be edited. The modem 6 controls communication via a telephone line. Here, in the modem 6, for example, an image or sound received from the Internet or the like can be a target of editing, and the edited image or sound can be transmitted to the outside.

  An I / O (Input / Output) interface 7 outputs an operation signal corresponding to the operation of the keyboard 8 and the mouse 9. The keyboard 8 is operated to input predetermined data and commands, and the mouse 9 is operated to move a cursor displayed on a display (computer display) 17 or to indicate a position. The

  The auxiliary storage interface 10 controls reading and writing of data with respect to a CD-R (Compact Disc Recodable) 11 and a hard disk (HD (Hard Disk)) 12. For example, an edited image or sound is recorded on the CD-R 11. The hard disk 12 stores an operating system, application programs for causing the microprocessor 1 to execute nonlinear editing and other processes. Furthermore, the image and sound to be edited, the edited image and sound, and the like are also recorded on the hard disk 12.

  The compression unit 13 compresses and encodes the image and sound input thereto in accordance with, for example, the MPEG standard. In the compression unit 13, data supplied via the expansion bus, data supplied via the decompression unit 15, and data supplied from an external device such as a video camera 14 are also included. It is made so that it can be compressed.

  In the video camera 14, for example, an image or sound to be edited is recorded. The compression unit 13 has an interface with the video camera 14, so that an image or sound recorded by the video camera 14 can be input to the compression unit 13.

  The decompression unit 15 decodes (decompresses) the data encoded (compressed) by the compression unit 13 and outputs the data. The decompressing unit 15 overlays and outputs the decoded image on the image stored in the frame buffer 3 as necessary. Here, the supply of image data from the frame buffer 3 to the decompression unit 15 is performed directly between them in the embodiment of FIG. 1, but in addition, the internal bus, the bus bridge 4, and It can also be done via an expansion bus. However, when the image data is supplied from the frame buffer 3 to the decompression unit 15 via the internal bus, the bus bridge 4 and the expansion bus, the data is congested if the capacity of the internal bus or the expansion bus is low. There is a fear.

  The VTR 16 records the image and sound output from the decompression unit 15 as necessary. The display 17 displays the image output from the decompression unit 15 as necessary. The image output from the decompression unit 15 can be displayed on a TV monitor or the like in addition to the display 17 that is a computer display.

  Next, FIG. 2 is a block diagram showing a functional configuration example of the editing apparatus of FIG. In the figure, portions corresponding to those in FIG. 11 are denoted by the same reference numerals, and description thereof will be omitted below as appropriate. That is, this editing apparatus is provided with a controller 24 (data accumulation amount calculation means) instead of the controller 124, and a bit rate estimation device 26 (bit rate estimation means) and a buffer accumulation amount adjustment device 27 (data amount adjustment). 11 is basically the same as that in FIG. 11 except that a means is newly provided.

  The controller 24 basically controls the data storage device 21, the bit counter 22, and the decoder 23 as well as the bit rate estimation device 26 and the buffer accumulation amount adjustment device 27 in the same manner as the controller 124 of FIG. 11. Has been made. Based on the bit amount output from the bit counter 22 and the information output from the controller 24, the bit rate estimation device 26 stores the bits of the bit streams A and B as the editing target material recorded in the data storage device 21. The rate is accurately estimated and output to the controller 24. The buffer accumulation amount adjusting device 27 is configured to adjust the data amount of the bit stream C as an editing result of the bit streams A and B.

  In FIG. 2, the data storage device 21 is in the auxiliary storage interface 10 and the hard disk 12 in FIG. 1, the bit counter 22 is in the microprocessor 1 in FIG. 1, the decoder 23 is in the decompression unit 15 in FIG. 1, the memory 25 corresponds to the main memory 2 in FIG. 1, and the bit rate estimation device 26 and the buffer accumulation amount adjustment device 27 correspond to the microprocessor 1 in FIG.

  Next, the operation will be described.

  Also here, for example, as shown in FIG. 3 similar to FIG. 12, a part of the scene A ′ is cut out from the bit stream A, and a part of the scene B ′ is cut out from the bit stream B. It is assumed that the bit stream C is configured by connecting the scene B ′ later. In addition, it is assumed that bit streams A and B that have been compression-encoded in accordance with the MPEG1 standard in the compression unit 13 are already recorded in the data storage device 21.

  The controller 24 reads the bit streams A and B from the data storage device 21 and supplies them to the bit counter 22 and further to the decoder 23. In the bit counter 22, the bit amount of the bit stream from the data storage device 21 is calculated and supplied to the bit rate estimation device 26. Further, the controller 24 supplies necessary information to the bit rate estimation device 26. The bit rate estimation device 26 accurately estimates the bit rates A and B based on the bit amount from the bit counter 22 and the information from the controller 24, and outputs the estimated bit rates to the controller 24.

  Here, details of the processing in the bit rate estimation device 26 will be described later.

  When the controller 24 receives the bit rate estimation result from the bit rate estimation device 26, in the following processing, when the bit rate is necessary, the controller 24 performs processing using the bit rate estimation result from the bit rate estimation device 26. .

  Thereafter, as shown in the flowchart of FIG. 4, the controller 24 performs the same processing in steps S <b> 1 to S <b> 17 as in steps S <b> 101 to S <b> 117 of FIG. 13.

Here, the determination as to whether or not the start point A _in is _in step S6 is performed as follows, for example. That is, as described above, the start point A _in can be set at the head of the GOP, and the GOP starts with the group start code GSC. In addition, a sequence header starting with a sequence header code SHC may be added before the GOP header. Therefore, first, SHC and GSC are detected.

When the start point A _in is specified by, for example, a time code, since the time code is arranged in the TC (Time Code) of the GOP header, the controller 24 receives the TC from the decoder 23, Based on this, it is determined whether or not the starting point is A _in . In addition, when the start point A _in is specified by, for example, the number of pictures from the first picture constituting the bit stream A, the controller 24 sends the picture start code PSC arranged in the picture header from the decoder 23. And a temporal reference TR (Temporal Reference) arranged thereafter is received. Since TR is reset in units of GOP and represents a so-called serial number of the pictures constituting the GOP, the number of pictures constituting the GOP can be recognized by obtaining the maximum value of TR in the GOP. Since TR starts from 0, the value obtained by adding 1 to the maximum value of TR is the number of pictures constituting the GOP. The controller 24 counts the number of pictures as described above, and determines whether or not the starting point is A _in based on the counted number.

The start point B _in and the end points A _out and B _out are similarly detected. However, when the end point A _out or B _out is set at the end of the bit stream A or B when the designation is made by the time code, the corresponding time code cannot be detected. . Therefore, if a time code that matches the time code corresponding to the end point A _out or B _out cannot be detected before the sequence end code SEC is detected, the time obtained in the last GOP When the addition value obtained by adding the time corresponding to the maximum value +1 of the temporal reference TR in the GOP to the code TC matches the time code corresponding to the end point A _out or B _out , the code TC immediately before the sequence end code SEC is added. The picture is set to end point A _out or B _out .

In the controller 24, the processing of steps S1 to S17 in FIG. 4 is performed. In step S14, when it is determined that the picture of the end point A _out has been read, the process proceeds to step S21 in FIG. The read data is recorded in the data storage device 21 as data about the bit stream C, and the process proceeds to step S22.

In step S22, the controller 24 calculates the data accumulation amount Buf (A _out ) of the VBV buffer at the end point A _out according to the following equation.

Buf (A _out ) = VD (A _out ) × R / 90000 + HL (A _out )
... (2)
However, in Expression (2), VD (A _out ) represents the VD of the picture at the end point A _out , and HL (A _out ) is the first bit of the sequence header or GOP header for the picture at the end point A _out. This represents the bit amount from (the first bit of the sequence header code SHC or the group start code GSC) to the last bit of the picture start code PSC. R represents an estimated value of the bit rate from the bit rate estimating device 26.

Note that when the end point A _out is set at the end of the bit stream A, VD (A _out ) cannot be obtained, so the data accumulation amount Buf (A _out ) of the VBV buffer at the end point A _out is It is calculated according to the following equation instead of equation (2).

Buf (A _out ) = Buf (A _out ′) −DL (A _out ′)
... (3)
However, in Expression (3), A _out ′ means the picture immediately before the picture of the end point A _out , and therefore Buf (A _out ′) is the picture immediately before the picture of the end point A _out . The amount of data stored in the VBV buffer in a picture, calculated according to equation (2). In addition, DL (A _out ′) is determined from the first bit of any one of the sequence header code SHC, the group start code GSC, or the picture start code PSC for the picture A _out ′. Represents the bit amount up to the bit immediately before any of the sequence header code SHC, the group start code GSC, or the picture start code PSC for the picture of the end point A _out and is obtained based on the output of the bit counter 22. .

After the calculation of the data accumulation amount Buf (A _out ) of the VBV buffer at the end point A _out, the same processing as in steps S 124 to S 127 of FIG. 14 is performed in steps S 23 to S 26.

Then, in step S26, if the picture at the start point B _in is determined to have been read, the process proceeds to step S27, the controller 24, the latest sequence header each other of the bit streams A and B are stored in the memory 25 And determine whether they match. If it is determined in step S27 that the latest sequence headers for the bit streams A and B do not match, the process proceeds to step S28, and the controller 24 updates the latest sequence for the bit stream B stored in the memory 25. The header is recorded in the data recording device 21 as a sequence header of the bit stream C, and the process proceeds to step S29. If it is determined in step S27 that the latest sequence headers for the bitstreams A and B match, step S28 is skipped and the process proceeds to step S29.

  Even when it is determined in step S27 that the latest sequence headers for the bit streams A and B match, the latest sequence header for the bit stream B stored in the memory 25 is stored in the bit stream. It may be recorded in the data recording device 21 as a C sequence header.

In step S29, the controller 24 calculates the data storage amount Buf (B _in ) of the VBV buffer at the start point B _in in the same manner as in step S22. The process proceeds to step S30 and the VBV buffer at the end point A _out is calculated. It is determined whether the data accumulation amount Buf (A _out ) matches the data accumulation amount Buf (B _in ) of the VBV buffer at the start point B _in .

If it is determined in step S30 that Buf (A _out ) and Buf (B _in ) match, that is, if the VD at the end point A _{out and} the VD at the start point B _in match. Since it is not necessary to perform processing for ensuring VD consistency at the connection point between the bit streams A and B in steps S31 to S36 described later, the processing is skipped and the process proceeds to step S41 in FIG.

If it is determined in step S30 that Buf (A _out ) and Buf (B _in ) do not match (normally they do not match as described above), the process proceeds to step S31, and VD matching is hereinafter performed. A process for adjusting the amount of data for the purpose of performance is performed.

That is, in step S31, the controller 24 determines whether Buf (A _out ) is greater than Buf (B _in ). If it is determined in step S31 that Buf (A _out ) is larger than Buf (B _in ), a larger amount of data is read from the VBV buffer by Buf (A _out ) −Buf (B _in ). Is performed.

  Specifically, in this case, the process proceeds to step S 32, and under the control of the controller 24, the stuffing code is recorded in the data storage device 21 as data constituting the bit stream C by the buffer accumulation amount adjusting device 27. The process proceeds to step S41 of FIG.

Here, the stuffing code means data having a value of 0 in bytes that can be arranged before the start code. In MPEG, the stuffing code is skipped at the time of decoding and does not affect the decoded image, but is counted as data constituting the bit stream in the management of the VBV buffer. Thus, the stuffing code, by recording as data of a picture at the end point A _out, when decoding picture at the end point A _out, it is possible to increase the amount of data read out from the VBV buffer, thereby, Buf (A _out) and Buf (B _in) and can be matched with.

  That is, in step S32, the stuffing code having the number of bytes C represented by the following equation is recorded.

C = INT ((Buf (A _out ) −Buf (B _in )) / 8)
... (4)
However, in the formula (4), INT () means the maximum integer equal to or less than the value in parentheses ().

On the other hand, in step S31, if Buf (A _out) is not greater than Buf (B _in), i.e., if Buf (A _out) is Buf (B _in) is smaller than, Buf (B _in) -Buf A process for increasing the data storage amount by (A _out ) is performed.

That is, in this case, the process proceeds to step S33, and under the control of the controller 24, the buffer accumulation amount adjusting device 27 converts the data about the skipped P picture into the data of the picture at the end point A _out as the data constituting the bit stream C. Recorded after.

  Here, a skipped P-picture is a so-called prediction error in which a motion vector is 0 ((0, 0)) and an I-picture or P-picture decoded in advance is decoded as a reference picture. Is a picture with 0, and its data amount is very small. For example, when a picture is composed of horizontal 352 pixels × vertical 240 pixels, the amount of data obtained by MPEG encoding a skipped P picture is 232 bits (29 bytes).

On the other hand, as described above with reference to FIG. 15, the data corresponding to the bit rate / picture rate is accumulated after the data of a certain picture is read out from the VBV buffer until the next picture is read out. The Therefore, by inserting a picture with a small amount of data such as a skipped P picture, the amount of data stored in the VBV buffer in which data is stored at a constant bit rate can be greatly increased. When the picture is added, the data storage amount Buf (P) of the VBV buffer can be set to Buf (B _in ) or more.

After the skipped P picture is inserted (recorded), the process proceeds to step S34, and the controller 24 calculates the data accumulation amount Buf (P) of the VBV buffer after the skipped P picture is inserted, and the process proceeds to step S35. When the data accumulation amount before insertion of the skipped P picture is Buf (P ₋₁ ), the data accumulation amount Buf (P) after the insertion is calculated according to the following equation.

Buf (P) = Buf (P ₋₁ ) + R / pr−D _p
... (5)
However, in Equation (5), R is the estimated bit rate obtained from the bit rate estimating device 26, pr is the picture rate obtained from the picture rate PR arranged in the sequence header, and D _p is It means the data amount of skipped P picture.

In step S35, the data storage amount of the VBV buffer after insertion of skipped P picture Buf (P) is, whether a Buf (B _in) or more, as determined by the controller 24. In step S35, if Buf (P) is determined not become Buf (B _in) or more, the process returns to step S33, further, is inserted skipped P picture, the data storage amount in the VBV buffer is further increased The

Further, in step S35, if Buf (P) is determined to become Buf (B _in) or more, the process proceeds to step S36, under the control of the controller 24, the buffer fullness adjustment device 27, as in step S32 Similarly, the stuffing code is recorded so that the data accumulation amount of the skipped P picture inserted last matches Buf (B _in ), and the process proceeds to step S41 in FIG.

Note that the stuffing code is recorded in step S36, in the case where Buf (P) is larger than Buf (B _in), if they are made to coincide, the process of step S36 is skipped The

  In step S41 of FIG. 6 and further subsequent steps S42 to S45, the same processing as in steps S133 to S137 of FIG. 14 is performed, and the editing process is terminated.

As described above, a bit stream (third bit stream) in which at least a part of bit streams A and B (first and second bit streams) obtained as a result of MPEG encoding is connected is configured by editing. In this case, the data accumulation amount of the VBV buffer at the end point A _out (first connection point) of the bit stream A and the start point B _in (second connection point) of the bit stream B is calculated based on VD. Since the data amount of the portion corresponding to the connection point of the bit streams A and B (A ′ and B ′) in the bit stream C is adjusted based on the difference between the data accumulation amounts, the connection point Can be matched without rewriting the VD of the bitstream B. As a result, the bitstreams A and B can be edited at high speed. Further, it is no longer necessary to MPEG-decode the bit streams A and B once.

In the above case, when Buf (A _out ) is smaller than Buf (B _in ), skipped P pictures are inserted and stuffing codes are recorded in steps S33 to S36, and data storage in the VBV buffer is performed. The amount is increased by Buf (B _in ) −Buf (A _out ), but this can also be performed as follows, for example.

That is, in order to increase the data accumulation amount of the VBV buffer by Buf (B _in ) −Buf (A _out ), the amount of data read from the VBV buffer at the time of decoding the picture at the end point A _out is changed to Buf (B _in ) -Buf (A _out ) should be reduced. Therefore, when a stuffing code exists as picture data of the end point A _out , this can be performed by deleting the stuffing code. However, in order to adjust the data accumulation amount by this method, the data amount of the stuffing code existing as the picture data of the end point A _out needs to be Buf (B _in ) −Buf (A _out ) or more. There is.

Further, after the end point A _out, when inserting the skipped P picture, at the time of decoding, so that the same image and the picture of the end point A _out is continuous. If such an identical image is present in the middle of a certain scene, the movement of the image becomes awkward and the viewer may feel uncomfortable. Here, however, such an identical image continues. Since this is a so-called editing point that is a connection point between the bit streams A and B, the awkwardness of the motion of the image is not conspicuous, and the viewer does not feel so uncomfortable.

  Next, the syntax of the skipped P picture inserted in step S33 of FIG. 5 is shown below. Here, it is assumed that the picture is composed of horizontal 352 pixels × vertical 240 pixels.

picture () {
picture_start_code 0x0000010
temporal_reference 10 bit unsigned int
picture_coding_type 010
vbv_delay (calculated by calculation)
full_pel_forward_vector 0
forward_f_code 001
extra_bit_picture 0
byte_alignment 000000
slice () {
slice_start_code 0x00000101
quantizer_scale 11111
extra_bit_slice 0
macroblock () {
macroblock_address_increment 1
macroblock_type 001
motion_horizontal_forward_code 1
motion_vertical_forward_code 1
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_escape 0000 0001 000
macroblock_address_increment 0000 0011 001
macroblock_type 001
motion_horizontal_forward_code 1
motion_vertical_forward_code 1
}
}
byte alinment 0
}

The temporal reference temporal_reference is a value obtained by adding 1 to the temporal reference temporal_reference of the I picture or P picture when the picture of the end point A _out is an I picture or P picture, and the picture of the end point A _out is In the case of a B picture, the value is set to a value obtained by adding 2 to the temporal reference of the B picture. That is, for example, as shown in FIG. 7A, when the picture at the end point A _out is a P picture and its temporal_reference is 19, the temporal_reference of the skipped P picture is set to 20, which is 19 plus 1. Also, for example, as shown in FIG. 4B, when the picture of the end point A _out is a B picture and its temporal_reference is 17, the temporal_reference of the skipped P picture is 19 which is 2 plus 2.

  The VBV delay vbv_delay is obtained by the following equation.

VD [N] = INT (VD [N-1] + 90000 × (1 / pr-BC [N-1, N] / R)
... (6)
In equation (6), VD [N] means the VBV delay of the Nth picture, and BC [N−1, N] is the data amount of the N−1th picture as shown in FIG. means. Note that BC [N−1, N] is a bit counter 22 that indicates, for example, the bit amount from the bit next to the picture start code PSC of the N−1th picture to the last bit of the start code PSC of the Nth picture. It shall be obtained by counting. A stuffing code may exist before the start code (start_code) arranged in each header, which is also included in the header length. Furthermore, pr represents a picture rate obtained as described later.

  In the above case, the macroblock_escape is repeated nine times. This is a case where the picture is composed of 352 horizontal pixels × 240 vertical pixels. In general, the macroblock_escape is expressed by the following equation. Repeated MEC times.

MEC = INT ((number of macroblocks of picture−1) / 33)
... (7)

  Furthermore, in the above-described case, macroblock_address_increment is set to 0000 0011 001, but this is also a case where a picture is composed of 352 horizontal pixels × 240 vertical pixels. Generally, macroblock_address_increment is a macroblock of a picture. The remainder when the number is divided by 33 is the value converted according to Table B.1 defined in ISO-11172-2.

  Next, a process for estimating a highly accurate bit rate from the bit rate BR arranged in the sequence header, which is performed in the bit rate estimating apparatus 26 of FIG. 2, will be described with reference to the flowchart of FIG.

In the bit rate estimating device 26, an accurate bit rate is estimated from the data amount between two points of the bit stream and the bit rate BR. First, these two points P ₁ and P ₁ are estimated. ₂ is set in step S51. As the point P ₁ or P ₂ , for example, a start point A _in or an end point A _out is set for the bit stream A, and a start point B _in or an end point B _out is set for the bit stream B, respectively. However, the points P ₁ and P ₂ may be set anywhere as long as their interval (number of pictures) satisfies a condition as described later.

Here, after the two points P ₁ and P ₂ are set, predetermined information is supplied from the bit counter 22 and the controller 24 to the bit rate estimation device 26.

That is, from the bit counter 22, for example, as shown in FIG. 10, the data amount BC [P ₁ from the bit next to the picture start code PSC at the point P ₁ to the last bit of the picture start code PSC at the point P ₂ , P ₂ ] is supplied. In FIG. 10, points P ₁ and P ₂ are set at positions where the sequence header SH and the GOP header GH are present, but the points P ₁ and P ₂ are the sequence header SH or the sequence header SH and the GOP header GH. It can be set even if there is no position.

Further, from the information decoded by the decoder 23, the controller 24 sends the picture rate PR and bit rate BR arranged in the sequence header, and the VBV delay VD [at points P ₁ and P ₂ arranged in the picture header. P ₁ ] and VD [P ₂ ] are supplied.

  In the bit rate estimation device 26, the picture rate PR and the bit rate BR are converted as follows. That is, according to the MPEG1 standard, the picture rate PR is 23.976 Hz, 24.000 Hz, 25.000 Hz, 29, 970 Hz, 30.000 Hz, 50.000 Hz, 59.940 Hz, Each is converted to 60.000 Hz to obtain a picture rate pr. The bit rate BR is converted to a bit rate br obtained by multiplying the bit rate BR by 400. Here, as described above, since BR is a value obtained by rounding up a value of 400 bps or less of the actual bit rate and dividing by 400, the actual bit rate is a value in the range of br-399 to br. become.

After receiving the necessary information from the bit counter 22 and the controller 24 after the processing of step S51, the bit rate estimation apparatus 26 proceeds to step S52, where the number of pictures n existing between the points P ₁ and P ₂ is candidate. Is calculated.

Here, the number n of pictures existing between the points P ₁ and P ₂ can be obtained according to the following equation.

n = ((VD [P ₂ ] −VD [P ₁ ] + E ₂ −E 1) / 90000
+ BC [P ₁ , P ₂ ] / br) × pr
... (8)
However, in Formula (8), E2 or E1 represents the rounding error of VD [P ₂ ] or VD [P ₁ ], and both are values of 0 or more and less than 1. Also, pr represents the picture rate obtained as described above from the picture rate PR. In addition, pr which is not an integer is expressed as follows. That is, when PR is 1, 4, or 7, the picture rate pr is set to 24000/1001, 30000/1001, or 60000/1001.

Since VD [P ₂ ] or VD [P ₁ ] has a rounding error E2 or E1, respectively, and the bit rate br also has a rounding error of less than 400 bps, the picture number n is obtained as a value within a certain range. However, since n must be an integer, it is limited to a certain number of candidates. In step S52, such candidates for the number of pictures n are obtained.

  Thereafter, in step S53, the bit rate estimation device 26 calculates a bit rate R (a candidate for an estimated bit rate value) using the candidates for the number of pictures n obtained in step S52.

  Here, the candidate of the bit rate R can be obtained according to the following equation modified by replacing br in Equation (8) with R.

R = BC [P ₁ , P ₂ ] / (n / pr
− (DVD [P ₂ , P ₁ ] + E _2-1 ) / 90000)
... (9)
However, in Expression (9), dVD [P ₂ , P ₁ ] represents VD [P ₂ ] −VD [P ₁ ], and E _2-1 represents E ₂ −E ₁ .

As described above, since there are a plurality of candidates for the picture number n, several candidates are also generated for the bit rate R. Further, since dVD [P ₂ , P ₁ ] has a rounding error, the number of candidates for the bit rate R is further expanded. However, since the bit rate R is a value in the range from br-399 to br obtained by multiplying BR by 400, it is limited to a certain number of candidates. In step S52, such a candidate for the bit rate R is obtained. Note that, due to this limitation, candidates for the number of pictures n are further limited.

  After the candidates for the number n of pictures and the candidates for the bit rate R are obtained as described above, the process proceeds to step S54, and a probable combination of the number n of pictures and the bit rate R is obtained from these candidates. It is done.

That is, in step S54, first, a third point P ₃ other than the points P ₁ and P ₂ is set on the bit stream. In this case, the VBV delay VD [P ₃ ] at the point P ₃ can be obtained according to the following equation.

VD [P ₃ ] = INT (VD [P ₁ ] + E1
+ 90000 × (n / pr−BC [P ₁ , P ₃ ] / R))
... (10)
However, in the equation (10), BC [P ₁ , P ₃ ] represents the data amount between the points P ₁ and P ₃ obtained in the same manner as described in FIG.

In step S54, the equation (10), together with substituting candidate of the candidate or bit rate R of a picture number n obtained respectively step S52 or S53, VBV delay at the point P _3, which is arranged in the actual bit stream Substituting VD [P ₃ ], a combination of the number of pictures n and the bit rate R that makes the difference between the right and left sides closer to 0 is obtained.

  Then, the process proceeds to step S55, and from the bit rate R obtained in step S54, the number N of pictures for which the error between the bit rate R and the true bit rate falls within a predetermined range is obtained.

  That is, now, when the equation (8) is modified by replacing the bit rate br with the true bit rate r, the following equation is obtained.

n = (dVD [P ₂ , P ₁ ] / 90000 + E _2-1 / 90000
+ BC [P ₁ , P ₂ ] / r) × pr
(11)

Assuming that dVD [P ₂ , P ₁ ] = 0, equation (11) becomes:

n = (E _2-1 / 90000 + BC [P ₁ , P ₂ ] / r) × pr
(12)

On the other hand, when the data amount BC [P ₁ , P ₂ ] is expressed by the estimated value R of the bit rate, it is as follows.

BC [P ₁ , P ₂ ] = n / pr × R
... (13)

  Substituting equation (13) into equation (12) yields:

n = (E _2-1 / 90000 + n / pr × R / r) × pr
= R * E < _2-1 > / (r-R) * 90000 * pr
(14)

Since | E _2-1 | is a value less than 1, n is the maximum value, and is expressed by the following equation.

n = r / (r−R) × 90000 × pr
... (15)

  For example, in the case of an NTSC image, Equation (15) is as follows.

n = r / ((r−R) × 3003.0)
... (16)

  In this case, the error in the bit rate estimation value R, that is, the number of pictures n for r−R to be 1 bps or less, for example, is r / 3003.

The above is a case where it is assumed that dVD [P ₂ , P ₁ ] = 0. For example, as in the case of VD [P ₂ ]> VD [P ₁ ], VD [P ₂ ] and VD [P ₁ ] Are not equal, the points P ₂ and P ₁ need to be set with an interval corresponding to the number of pictures equal to or greater than dVD [P ₂ , P ₁ ] / 90000 × pr.

Now, a bit rate having an error is represented by r ′, and the error of the bit rate r ′ affects the calculation of the VBV delay VD [A _out ] and VD [B _in ] at the end point A _out and the start point B _in . Is considered, the following equation is obtained from the equation (10).

E _VBV = 90000 × BC [P ₁ , P ₂ ] × (r′−r) / r × r ′
... (17)
However, in Expression (17), E _VBV represents an error of the VBV delay.

For simplification, BC [P ₁ , P ₂ ] = r × t (where t represents the time corresponding to the interval between the points P ₁ and P ₂ ), the equation (17) Can be modified as follows.

t = 1 / (r′−r) × E _VBV / 90000 × r ′
... (18)

Since E _VBV is rounded to an integer, its minimum value is 1. Substituting this into equation (18) yields:

t = 1 / (r′−r) × r ′ / 90000
... (19)

  Therefore, if the bit rate error, i.e., r'-r is, for example, 1 bps, the calculation of the VBV delay is affected when the time t is shorter than the time represented by the following equation. become.

t = r '/ 90000
... (20)

Therefore, for example, in the case of a video CD (Compact Disc), the bit rate is about 1.15 Mbps. In this case, in order to set the bit rate error to 1 bps, the interval between the points P ₁ and P ₂ is set. About 12.8 seconds or more. In this case, for example, in order to set the bit rate error to 2 bps, the interval between the points P ₁ and P ₂ needs to be about 6.4 seconds or more, which is half the above-mentioned.

Therefore, if the points P ₁ and P ₂ are set at such an interval or more, the error of the estimated bit rate R obtained in step S54 becomes so small that the influence can be ignored.

Note that, as described above, only the distance between the points P ₁ and P ₂ and the error of the bit rate are taken into consideration, but in the equation (10) used in step S54, the VBV delay VD [P ₁ ] at the point P ₁ is used. Rounding error, and the effect of this error is greater.

  In step S55, as described above, the number N of pictures corresponding to a time during which the error of the bit rate R obtained in step S54 can be ignored is obtained.

Then, the process proceeds to step S56, and it is determined whether or not the number of pictures n between the points P ₁ and P ₂ obtained in step S54 is equal to or greater than the number N of pictures obtained in step S55. In step S56, if the number of pictures (n) is determined to be not more than N, the process proceeds to step S57, the like number of pictures between the points P ₁ and P ₂ is greater than or equal to N, is the point P ₁ and P ₂ Will be reset. And it progresses to step S52 and the above-mentioned process is repeated.

  On the other hand, if it is determined in step S56 that the number of pictures n is N or more, the estimated bit rate R obtained in step S55 is output to the controller 24, and the process is terminated.

  The controller 24 performs the above-described processing using the bit rate estimation value R thus obtained.

  As described above, since the bit rate is accurately estimated, the processing accuracy in the controller 24 can be improved.

  Note that the bit rate estimation device 26 calculates probable candidates (estimated values) of the respective bit rates for both the bit streams A and B as described above and supplies them to the controller 24. Then, the controller 24 uses the bit rate candidates of both the bit streams A and B, for example, the average value of the overlapping portions, for performing the above-described processing.

  However, if the bit rate candidates of both bit streams A and B do not have overlapping portions and there is a significant difference, the bit rates at the time of encoding are considered to be different. In this case, editing cannot be performed because of the MPEG1 standard, and the editing process is stopped.

  The case where the present invention is applied to an editing apparatus that edits a bitstream obtained by performing encoding in accordance with the MPEG1 standard has been described above. However, the present invention is not limited to MPEG1 encoding but MPEG2 encoding. It can be applied to the case of editing a bitstream obtained by encoding or performing encoding assuming a virtual buffer on the other receiving side.

  The present invention can also be applied to editing a bitstream including, for example, audio in addition to an image.

It is a block diagram which shows the structural example of one Embodiment of the editing apparatus to which this invention is applied. It is a block diagram which shows the functional structural example of the editing apparatus of FIG. It is a figure which shows a mode that the bit stream A and B are edited and the bit stream C is comprised. 3 is a flowchart for explaining the operation of the editing apparatus in FIG. 2. It is a flowchart following the flowchart of FIG. It is a flowchart following the flowchart of FIG. It is a figure for demonstrating how to attach the temporal reference at the time of inserting a skipped P picture. It is a figure for demonstrating how to obtain | require the VBV delay of a skipped P picture. It is a flowchart for demonstrating operation | movement of the bit rate estimation apparatus 26 of FIG. It is a figure for demonstrating how to obtain | require the estimated value of a bit rate. It is a block diagram which shows the structure of an example of the conventional editing apparatus. It is a figure which shows a mode that bit stream A and B are edited and bit stream C is comprised. 12 is a flowchart for explaining the operation of the editing apparatus of FIG. 11. It is a flowchart following the flowchart of FIG. It is a figure which shows transition of the data storage amount of a VBV buffer. It is a figure which shows a mode that inconsistency arises in the data storage amount of a VBV buffer when bit stream A and B are connected.

Explanation of symbols

  1 Microprocessor, 2 Main memory, 3 Frame buffer, 4 Bus bridge, 5 Tuner, 6 Modem, 7 I / O interface, 8 Keyboard, 9 Mouse, 10 Auxiliary storage interface, 11 CD-R, 12 Hard disk, 13 Compression unit , 14 video camera, 15 expansion unit, 16 VTR, 17 computer display, 21 data storage device, 22 bit counter, 23 decoder, 24 controller (data storage amount calculation means), 25 memory, 26 bit rate estimation device (bit rate estimation) 27) Buffer accumulation amount adjustment device (data amount adjustment means)

Claims (9)

Editing at least the first and second bitstreams obtained by performing compression coding on the image, and forming a third bitstream connecting at least a part of each of the first and second bitstreams An image editing device,
In the case where the compression encoding is performed in consideration of the data accumulation amount of the virtual buffer assumed on the receiving side, and accumulation amount information regarding the data accumulation amount is included in the first and second bitstreams,
The virtual buffer at the first connection point of the first bit stream connected to the second bit stream and the second connection point of the second bit stream connected to the first bit stream. Data storage amount calculation means for calculating the data storage amount based on the storage amount information;
The connection between the first and second bit streams in the third bit stream based on the difference between the data accumulation amounts at the first and second connection points calculated by the data accumulation amount calculating means. An image editing apparatus comprising: a data amount adjusting unit that adjusts a data amount of a portion corresponding to a point. The first and second bitstreams are at least images that are compression-encoded in accordance with MPEG (Moving Picture Experts Group) standards,
The image editing apparatus according to claim 1, wherein the accumulation amount information is a VD (VBV Delay) included in a picture header corresponding to a data accumulation amount of a VBV (Video Buffering Verifier) buffer. The data amount adjusting means inserts a stuffing code into the picture data at the first connection point when the data accumulation amount at the first connection point is larger than the data accumulation amount at the second connection point. The image editing apparatus according to claim 2, wherein: The image editing apparatus according to claim 3, wherein the data amount adjusting unit inserts the stuffing code so that data accumulation amounts at the first and second connection points are equal. When the data accumulation amount at the first connection point is smaller than the data accumulation amount at the second connection point, the data amount adjusting means skips a skipped P picture (Skipped) after the picture data at the first connection point. The image editing apparatus according to claim 2, wherein data about P-picture is inserted. The data amount adjusting means inserts the skipped P picture until the data accumulation amount after inserting the skipped P picture becomes equal to or greater than the data accumulation amount at the second connection point. 5. The image editing apparatus according to 5. The data amount adjusting means inserts the skipped P picture so that when the data accumulation amount after the insertion becomes larger than the data accumulation amount at the second connection point, the data accumulation amount becomes equal. The image editing apparatus according to claim 6, wherein a stuffing code is inserted as described above. Bit rate estimating means for estimating a bit rate of the first and second bit streams,
The data accumulation amount calculation unit calculates a data accumulation amount of the VBV buffer based on the bit rate estimated by the bit rate estimation unit and the accumulation amount information VD. Image editing device. Editing at least the first and second bitstreams obtained by performing compression coding on the image, and forming a third bitstream connecting at least a part of each of the first and second bitstreams An image editing method,
In the case where the compression encoding is performed in consideration of the data accumulation amount of the virtual buffer assumed on the receiving side, and accumulation amount information regarding the data accumulation amount is included in the first and second bitstreams,
The virtual buffer at the first connection point of the first bit stream connected to the second bit stream and the second connection point of the second bit stream connected to the first bit stream. The data storage amount of is calculated based on the storage amount information,
Based on the difference between the data accumulation amounts at the first and second connection points, the data amount of the portion corresponding to the connection point between the first and second bit streams in the third bit stream is adjusted. An image editing method characterized by:

Images