CN1136721C - Stream editing device and stream editing method - Google Patents

Abstract

The present invention relates to a flow editing system which is designed at an editing point arranged in each flow and is switched from one input flow to another input flow used during editing operation. The system comprises two decoders and at least one encoder. In the editing operation for merging two flows to generate one flow, the system only recodes the tail part of one flow and the front part of another flow which cross the editing point and uses the rest parts of the flows without change. Thus, the deterioration of signal quality caused by recoding the flows is lowered to minimum.

Description

Stream editing device and stream editing method

The present invention relates generally to a stream editing device for editing a stream composed of video data codes and audio data codes; in particular the present invention relates to a stream editing device for connecting two streams in real time , resulting in a single stream without image degradation.

The MPEG format is a typical video and audio code compression coding technology. In this format, discrete cosine transform (DTC) and predictive coding methods are used to reduce the amount of data, thereby realizing compression coding of moving images.

Figure 11 shows an encoder which is used to perform this encoding operation.

The encoder includes a subtractor 41 , a DCT circuit 42 , a quantizer 43 , a variable length coding circuit 49 , a buffer 50 , an inverse quantizer 44 , an inverse DCT circuit 45 , an adder 46 , an image memory 47 and a motion vector detector 48 .

When performing interframe predictive coding, the motion vector detector 48 compares the input image data with the image data stored in the image memory 47, calculates the motion vector (MV), and outputs it to the image memory 47 and variable length coding Circuit 49.

The subtractor 41 reads out the motion-compensated image data from the image memory 47 , and outputs the difference between the input image data and the image data read from the image memory 47 to the DCT circuit 42 . The DCT circuit 42 performs discrete cosine transform on the difference data input from the subtractor 41 , outputs its DCT coefficients to the quantizer 43 , and notifies the variable length coding circuit 49 of the type of DCT.

The quantizer 43 quantizes the DCT coefficients in a quantization step specified by the quantizer matrix (Quant) of the buffer 50 , and outputs the quantization result and the quantizer matrix (Quant) to the variable length encoding circuit 49 . The quantization result is also output to the inverse quantization circuit 44 .

The variable length encoding circuit 49 encodes the motion vector (MV) calculated by the motion vector detector 48 , the type of DCT, the quantizer matrix (Quant), and the output of the quantizer 43 . These encoded data are temporarily stored in the buffer 50, and then output in the form of a stream.

The output of the quantizer 43 is also inversely quantized by an inverse quantizer 44 and then inversely DCTed in an inverse DCT circuit 45 . The adder 46 adds the output of the inverse DCT circuit 45 to the image data read out from the image memory 47, reproduces the input image data, and then stores the image data in the image memory 47 for the next input image data Perform a subtraction operation.

When inter-frame encoding is performed without using inter-frame prediction, only DCT is performed, and image data is not read out from the image memory 47 .

When inter-frame prediction is used, it is also possible to perform forward inter-frame predictive coding using a previous picture stored in the picture memory 47 and perform backward inter-frame predictive coding using future pictures stored in the picture memory 47 .

In the MPEG format, each input frame is encoded as one of the following three pictures: an I picture containing only intra-macroblocks; a P picture containing intra-macroblocks with forward inter-frame prediction coding. pictures; and B-pictures containing intra, forward inter-prediction coded and reverse inter-prediction coded macroblocks.

10(a)-10(d) show a sequence of processes for adding audio data to video data encoded in the above-described method to generate a data stream in a format suitable for a storage medium.

Moving pictures are placed in a different frame order than when input and encoded into I pictures, P pictures, and B pictures to generate a video elementary stream (ES). Audio data is compressed, for example, at intervals of 24 milliseconds to generate audio ES. These ESs are divided into appropriate units and will be grouped together with the header. For each header, a PTS (representing a time stamp) is added, and the PTS represents the time at which the packet data is reproduced.

The video and audio packets thus formed are multiplexed together with a header as one pack to form a stream. Figure 9 shows a typical compressed stream. Record the stream on a storage medium such as DVD. The stream is read out from the storage medium, and each packet is decoded and reproduced at the timing specified by the PTS, thereby realizing playback.

Generally, a stream having a desired content is generated from a plurality of streams recorded in a storage medium by reading out the streams from the storage medium and rearranging them. For example, when stream A needs to be split into two parts and stream B is placed between them to produce stream C, the first part of stream A is decoded and will be re-encoded to form the front of stream C. The second part of stream A is then decoded and re-encoded to produce the tail of stream C. However, such stream editing has a disadvantage in that repeated encoding operations degrade the quality of edited images.

Editing techniques that directly couple the streams have also been proposed, but they are limited by having to couple the streams at the I-picture of the video data.

In the MPEG format, video data is grouped in units of a group of pictures (GOP), where a GOP is composed of one I picture and a plurality of P and B pictures. Each GOP typically includes 15 pictures (0.5 seconds). If a stream is divided at one picture of a P picture or a B picture, it is difficult to reproduce a frame picture in units of pictures, thus causing a problem that the picture continues to be distorted until the next I picture appears.

The above edits are usually done while operating offline, so it is time-consuming.

Therefore, one main object of the present invention is to avoid the disadvantages of the prior art.

Another object of the present invention is to provide a stream editing system which can connect two streams in real time to produce a single stream without deteriorating signal quality such as image quality.

According to an aspect of the present invention, there is provided a stream editing device designed to switch one of two input streams used in editing operations to the other at an editing point set in each stream. The system includes (a) a first decoder for decoding a first stream; (b) a second decoder for decoding a second stream; (c) an encoder for decoding at least one of the decoded first and second streams is re-encoded; and (d) a controller for controlling editing of the first and second streams to produce a third stream consisting of two parts, One part is the front segment of the first stream, which is located before the edit point set in the first stream, and the other part is the tail segment of the second stream, which is located after the edit point set in the second stream, and the controller generates the third stream , at least one segment portion of the preceding and following segments in the first and second streams is merged with another segment portion of the preceding and following segments in the first and second streams, wherein at least one segment Segment parts are decoded by a corresponding one of the first and second decoders, the encoder re-encodes and are defined to leave the edit point set in one of the first and second streams by a given length, while the other segment Part refers to the part of the segment before it is decoded by the first and second decoder and re-encoded by the encoder.

In a preferred mode of the invention, both the first and second streams consist of groups of pictures (GOPs). The controller determines the length of the portion, which is decoded and re-encoded, and which will be merged in the third stream, from the beginning of the edit point to the front of one GOP where the edit point is set.

There may also be a header generator for generating a series of headers for the third stream.

The encoder may re-encode at least one of the first and second streams using encoding information obtained by decoding at least one of the first and second streams by the first and second decoders.

The encoding information includes image type, quantizer matrix (Quant) and discrete cosine transform (DCT) type.

When an encoder is required to change the bit rate used when encoding with one of the first and second streams when performing a re-encoding operation on at least one of the first and second streams, the encoder uses a change according to the relationship Quant:

Quant after change = Quant before change × (bit rate after change / bit rate before change)

The decoded video and/audio data and encoding information are transferred from the first and second decoders arranged in separate units to the encoder via two separate signal lines.

The device may also include demultiplexing means for dividing the first and second streams into video packets and audio packets; and multiplexing means for video packetizing one of the first and second streams and audio packets in another stream for multiplexing.

When the part to be merged into the third stream is defined in the trailer section of the first stream, the encoder edits the part after finishing editing the second stream.

In the process of generating the third stream, the controller selects the following four parts, one is the first part of the previous segment in the first stream before decoding and re-encoding, and the other is the first part after the first part, before the edit point, by the encoder The second part of the first segment in the first stream output, the third after the edit point, the first part of the last segment in the second stream output by the encoder, and the fourth after the first part of the second stream but before decoding and re-encoding The second part of the tail segment in the second stream. The controller determines the range of the second part of the first stream from the edit point to the front of a GOP in which the edit point is set, and the range of the first part of the second stream from the edit point to the end of a GOP in which the edit point is set.

According to a second aspect of the present invention, there is provided a stream editing method capable of switching one of two input streams used in an editing operation to the other at an edit point set in each input stream. The method comprises the steps of: (a) decoding the first and second streams with two decoders; (b) re-encoding for at least one of the first and second streams decoded in the decoding step; and (c ) controls the editing of the first and second streams so as to generate a third stream, the third stream consists of two parts, one part is the front segment of the first stream, which is located before the edit point set in the first stream, and the other part is the first segment The tail segment of the second stream, which is located after the edit point set in the second stream, the control step combines at least one segment portion of the preceding segment and the following segment in the first and second streams with the third stream when generating the third stream Another segment part merge of the preceding segment and the following segment in the first and second streams, where at least one segment part is decoded, re-encoded and defined away from an edit point set in one of the first and second streams A segment of a given length, and another segmented part refers to the segmented part before decoding and re-encoding.

In a preferred mode of the invention, both the first and second streams consist of groups of pictures (GOPs). The length of the part to be decoded and re-encoded and to be merged in the third stream is defined as the beginning of one GOP from the edit point to the set edit point.

The method may also comprise the step of generating a series of headers for the third stream.

The encoding step re-encodes at least one of the first and second streams using encoding information obtained by decoding at least one of the first and second streams in the decoding step.

The encoding information includes image type, quantizer matrix (Quant) and discrete cosine transform (DCT) type.

When the encoding step is required to vary the bit rate used in encoding one of the first and second streams, the encoding step uses Quant varied according to the following relationship:

Quant after change = Quant before change × (bit rate after change / bit rate before change)

The method may further comprise the steps of: separating the first and second streams into video packets and audio packets; and multiplexing video packets from one of the first and second streams and audio packets from the other stream.

When the part to be merged into the third stream is defined in the trailer section of the first stream, the encoding step edits the part after editing of the second stream is completed.

During the generation of the third stream, the control step selects the following four parts, one is the first part of the preceding segment in the first stream before decoding and re-encoding, the other is after the first part, before the edit point, the decoded and The second part of the pre-segment in the re-encoded first stream, the third is after the edit point, the first part of the tail segment in the second stream, decoded and re-encoded, the fourth is after the first part of the second stream, decoded and re-encoded The second part of the trailer segment in the second stream before encoding. The range of the second part of the first stream from the edit point to the front of one GOP where the edit point is set is determined. The range of the first part of the second stream from the edit point to the end of one GOP in which the edit point is set is determined.

With reference to the accompanying drawings of the preferred embodiments of the present invention and reading the following detailed description, the present invention will be more fully understood, however, the accompanying drawings and descriptions should not be considered as limiting the invention to the scope of the specific embodiments, they are only for the purpose of Ease of explanation and understanding.

Attached are:

FIG. 1 is a block diagram showing an editing system according to a first embodiment of the present invention;

Figure 2 is stream 0 and stream 1 edited with the editing system, and stream 3 composed of edited stream 0 and stream 1;

Fig. 3 is a block diagram showing an editing system according to a second embodiment of the present invention;

Fig. 4 is a block diagram, has shown the circuit structure of multiplexer in Fig. 3 editing system;

Fig. 5 is a block diagram showing an editing system according to a third embodiment of the present invention;

Fig. 6 is a graph showing the signal-to-noise ratio curves in three cases, the first case using all the same parameters as in the previous encoding operation, and the second case using only the same parameters as in the previous encoding operation The same parameters representing the image type, quantizer matrix (Quant) and DCT type, while the third case does not use the same parameters at all;

FIG. 7 is a block diagram illustrating an editing system according to a fourth embodiment of the present invention;

Fig. 8 is stream 0 and stream 1 edited with the editing system of the fourth embodiment in Fig. 7, and stream 3 made up of stream 0 and stream 1;

Figure 9 shows a compressed stream;

Figures 10(a)-10(d) show a sequence of processes for adding audio data to video data to produce a data stream in a format suitable for storage media;

Fig. 11 is a block diagram showing a conventional editing system.

Referring now to the drawings, like numerals indicate like parts throughout the several views. FIG. 1 shows an editing system according to the present invention which edits streams read out from a DVD-RAM 11 and re-stores them in the DVD-RAM 11. As shown in FIG.

The editing system generally includes a stream controller 12 for controlling reading and writing streams to DVD-RAM 11; two AV decoders 20 and 21 for decoding streams to generate video signals and audio signals; video encoder 18, For video signal re-encoding, to produce video elementary stream (ET); Audio encoder 19, for audio signal re-encoding, to produce audio elementary stream; Multiplexer 17, for video elementary stream and Multiplexing of audio elementary streams to produce compressed streams; delay circuit 15; AV encoder 22 for decoding audio and video signals to visually monitor them; switch 24 for video signals to be input Select to video encoder 18; switch 23 for selecting the audio signal to be input to audio encoder 19; switch 16 for selecting a stream from the streams output by multiplexer 17 and delay circuit 15; timing control 13 for controlling the operation of the switches 16, 23 and 24; and a system controller 14 for controlling the overall system operation.

Fig. 2 shows streams processed by the editing system of this embodiment.

The editing system combines the segments A and B portions of stream 0 with segments C and D of stream 1 to produce stream 3 . In each of stream 0 and stream 1, an edit point is set between segment B and segment C. In the following discussion, it is assumed that information on editing points is input to the system controller 14 by an editor or other external device before starting an editing operation. Streams 0, 1 and 3 are divided into smaller parts at regular intervals, each smaller part forming a group of pictures (GOP). The front of each GOP contains I pictures. In the illustrated case, edit points in each stream are set in a GOP a given interval from the front of the GOP. Segment B is defined from the edit point to the front of the GOP in which the edit point is set. Segment A consists of the GOP preceding Segment B. Segment C is defined from the edit point to the end of the GOP in which the edit point is set in stream 1 . Segment D is composed of GOPs that follow Segment C.

When editing operation is performed to generate stream 3, segment A of stream 0 read from DVD-RAM 11 is used as it is, while segment B is decoded by AV decoder 20 and passed through video encoder 18 and audio Encoder 19 is used after re-encoding. Section C of stream 1 read from DVD-RAM 11 is used after being decoded by AV decoder 21 and re-encoded by video encoder 18 and audio encoder 19, while section D is used as it is. Specifically, among the four segments A, B, C and D in the process of generating stream 3, only segments B and C are used after being re-encoded. This can alleviate the degradation of the image quality of stream 3 .

The stream controller 12 reads stream 0 and stream 1 from the DVD-RAM 11, and outputs it according to the timing at which the edit point in stream 0 coincides with the edit point in stream 1. The AV decoder 20 decodes the stream 0 output from the stream controller 12 , and the AV decoder 21 decodes the stream 1 output from the stream controller 12 .

The timing controller 13 operates the switch 16 in response to control signals sequentially output from the system controller.

First, the timing controller 13 operates the switch 16 to pass the segment A of the stream 0 output from the delay circuit 15 to the stream controller 12 . Specifically, the segment A of the stream 0 read out from the DVD-RAM 11 by the stream controller 12 undergoes a delay in the delay circuit 15 for a period of time, and then returns to the DVD-RAM 11 as it is.

Next, timing controller 13 operates switch 16 to establish communication between stream controller 12 and multiplexer 17, and also operates switch 23 and switch 24 to establish communication between audio encoder 19 and AV decoder 20 and video encoding. Communication is established between the device 18 and the AV decoder 20. Specifically, the video data and audio data in segment B of stream 0 are re-encoded by the video encoder 18 and audio encoder 19 after being decoded by the AV decoder 20 .

As described above, the video encoder 18 and the audio encoder 19 generate video elementary streams and audio elementary streams, respectively. These elementary streams are multiplexed in the multiplexer 17 to generate a compressed stream, and then the compressed stream is delivered to the stream controller 12 through the switch 16, and it is recorded in the DVD-RAM 11, following Behind segment A of flow 0. Delay circuit 15 delays the data in section A for a time equal to the time required to decode, encode and multiplex the data in section B.

Then, the timing controller 13 selects the communication between the audio encoder 19 and the AV decoder 21 and between the video decoder 18 and the AV decoder 21 through the switch 23 and the switch 24, and divides the stream 1 output by the AV decoder 21 The video data and audio data in section C are re-encoded. In the multiplexer 17, the video elementary stream and the audio elementary stream output by the video encoder 18 and the audio encoder 19 are multiplexed to generate a compressed stream, and then the compressed stream is delivered to the stream controller through the switch 16 12, and record it in DVD-RAM 11, following segment B of stream 0.

Finally, the timing controller 13 operates the switch 16 to pass the segment D of the stream 1 output by the delay circuit 15 directly to the stream controller 12 . Specifically, the segment D of the stream 1 read out from the DVD-RAM 11 by the stream controller 12 experiences a delay in the delay circuit 15 for a period of time, and then is recorded in the DVD-RAM 11 again, following the segment C of the stream 1. later. Delay circuit 15 delays the data in section D for a period of time equal to the time required to decode, encode and multiplex the data in section C.

The stream can be edited in real time using the two decoders 20 and 21 . During the generation of stream 3, only the segments B and C re-encoded by the encoders 18 and 19 can be used to couple the two streams 0 and 1 to each other without much degradation in image quality.

Section A of stream 0 cannot be rewritten into DVD-RAM 11. In this case, the re-encoded sequence of segments B and C is recorded on an area of DVD-RAM 11, for example, following the area where segment A was originally recorded, while segment D is copied on an area of DVD-RAM 11 an appropriate area.

In the above discussion, it is assumed that the edit points of streams 0 and 1 are both set in one GOP, but it is possible that one of the edit points is just set between two adjacent GOPs. For example, as shown in FIG. 2, if the edit point of stream 0 is set at the end of the latest GOP of segment A, then when generating stream 3, the part of stream 1 following segment B is used. Specifically, in generating stream 3, stream 0 before re-encoding is used as it is, and only the part of stream 1 from its edit point to the end of a certain GOP in which the edit point is set is re-encoded for reuse.

An editing system of a second embodiment for re-editing a header signal that can correctly decode a combined stream with a simple decoder will be described below.

In the first embodiment, stream 0 and stream 1 are coupled together to generate stream 3, which will cause the PTS and frame number contained in the packet header between stream 1 and stream 0 to be discontinuous. Therefore, when decoding stream 3, the decoder needs to analyze the discontinuous PTS and frame numbers in order to reproduce images in the correct order and to synchronize the reproduced sound with the images.

As shown in FIG. 3 , the editing system of this embodiment includes a demultiplexer 25 , a multiplexer 26 , and switches 27 and 28 . The demultiplexer 25 separates the stream 1 output from the stream controller 12 into video packets and audio packets. Multiplexer 26 reconnects the headers with the outgoing video packets and audio packets. The delay circuit 15 is connected between the switch 16 and the signal line extending from the stream controller 12 to the AV decoder 20 . The other structures are the same as those in the first embodiment shown in FIG. 1, so a detailed description thereof is omitted here.

As shown in Figure 4, multiplexer 26 comprises video packet header generator 32, is used to generate video packet header; Video packet generator 31, it produces video with the header that video packet header generator 32 produces Grouping; audio packet header generator 34, for generating audio packet header; audio packet generator 33, it generates audio packet with the header that audio packet header generator 34 produces; compression header generator 35, for Produce compression header; Output selector 36, be used for determining the type of grouping that will be multiplexed into a compression; One of the outputs of the audio packet generator 33 and the compression header generator 35 is selected.

As in the first embodiment, segment A of flow 0 is directly input to flow controller 12 through delay circuit 15 and switch 16 .

Next, segment B of stream 0 and segment C of stream 1 are merged after segment A. Specifically, timing controller 13 operates switches 27 and 28 to establish communications between multiplexer 26 and video encoder 18 and between multiplexer 26 and audio encoder 19 . Thus, the video elementary stream generated by the video encoder 18 and the audio elementary stream generated by the audio encoder 19 are supplied to the multiplexer 26 .

The video packet header generator 32 of the multiplexer 26 generates a packet header for the incoming video elementary stream. The video packet generator 31 separates the video elementary stream into video packets, and adds the packet header generated by the video packet header generator 32 to them, thereby outputting the video packet to the switch 37 together with the header.

Likewise, the audio packet header generator 34 generates a packet header for the input audio elementary stream. The audio packet generator 33 separates the audio elementary stream into audio packets, and adds the packet header generated by the audio packet header generator 34 to them, thereby outputting the audio packets together with the header to the switch 37 .

Compressed header generator 35 generates compressed headers.

The output selector 36 monitors the output from the switch 37 to the stream controller 12 while simulating the buffering operation of a preselected ideal decoder, and operates the switch 37, from the video packet generator 31, the audio packet generator 33 and the compressed header One of the outputs of the generator 35 is selected to follow the output to the stream controller 12 so that a compressed stream such as that shown in FIG. 9 is sent to the stream controller 12 .

Demultiplexer 25 separates segment D of stream 1 into video packets and audio packets, and outputs them to multiplexer 26 through switches 27 and 28, respectively.

The video packet header generator 32 generates a header for the video packet input from the demultiplexer 25 . The video packet generator 31 replaces the header of each input video packet with one of the headers output from the video packet header generator 32 to output the header-substituted video packet.

Also, the audio packet header generator 34 generates a header for the audio packet input to the demultiplexer 25 . The audio packet generator 33 replaces the header of each input audio packet with one of the headers output from the audio packet header generator 34, so as to output the header-substituted audio packet.

As above, output selector 36 determines the type of packets to be included in a compression, and operates switch 37 to generate a compressed stream.

As can be seen from the above discussion, the editing system of the present embodiment reconnects the header with the packets in section D and sections B and C. Therefore, a complete stream can be formed by increasing the number of frames and the PTS continuing from the header of section A of stream 0 to the headers of sections B, C, and D.

The editing system of the third embodiment, which can minimize the deterioration of image quality caused by the re-encoding operation, will be described below.

Fig. 5 shows the editing system of the present embodiment, and the difference between this system and the second embodiment in Fig. 3 is only that the system controller 14 takes out the parameters used when decoding from the decoders 20, 21 and 22, to Controls the operation of encoders 18 and 19. Other structures and operations are the same, so detailed descriptions thereof are omitted here.

As discussed in FIG. 11 , motion pictures can generally be coded with parameters such as picture type, quantizer matrix (Quant), DCT type, motion vector (MV) and macroblock type (MB). The encoder side sends the information of these parameters to the decoder side. The decoder side uses the received information to decode the input data.

The picture type represents one of I picture, P picture, and B picture, and is expressed in two bits per frame.

The quantizer matrix (Quant) consists of parameters, each parameter specifies the quantization step size, and is expressed in the form of five bits per macroblock (MB).

The type of DCT indicates whether the DCT is a frame DCT or a field DCT, and is expressed in the form of one bit per MB.

A motion vector (MV) is expressed in six or more bits per MB.

The type of MB indicates the encoding mode of the MB, and is expressed in the form of one to eight bits per MB.

It is possible to reduce the degradation of image quality that occurs when the video data decoded by the decoders 20 and 21 is re-encoded by the video encoder 18 by reusing the parameters used in the encoding operation between the decoding operations of the decoders 20 and 21 . This is because, again using the same parameters in the video encoder 18, the same encoding operation that the previous video data was subjected to can be performed.

Specifically, the system controller 14 retrieves the information of the above parameters decoded in the decoders 20 , 21 and 22 and sends it to the video encoder 18 . The video encoder 18 generates the same parameters as those used in the encoding operation that the input video data has previously undergone, based on the received information, and uses them to re-encode the video data, thereby mitigating the degradation of image quality.

The video encoder 18 may only use the picture type, quantizer matrix (Quant) and DCT type information.

Fig. 6 is a graph showing the signal-to-noise ratio curves in three cases, the first case using all the same parameters as in the previous encoding operation, and the second case using only the same parameters as in the previous encoding operation The same parameters representing image type, quantizer matrix (Quant) and DCT type, while the third case does not use the same parameters at all.

These curves illustrate that using only the same parameters representing the image type, quantizer matrix (Quant) and DCT type as in the previous encoding operation, it is possible to substantially achieve the signal-to-noise ratio (SNR) when all using the same parameters. In addition, these three parameters are smaller than the remaining parameters indicating MV and MB types in terms of the amount of data used, so information can be efficiently sent from the system control 14 to the video encoder 18 .

Generally, it is necessary to change the bit rate in the encoding operation before the decoding operation (hereinafter referred to as the original bit rate) at the time of the re-encoding operation. For example, joining two streams with different bit rates requires re-encoding one stream to match the bit rate of the other stream. When this happens, the editing system of this embodiment will modify each value of the quantizer matrix (original Quant) to generate a new quantizer matrix (new Quant) according to the following relationship:

Original Bit Rate: New Bit Rate = Original Quant: New Quant

Therefore, new Quant = original Quant × (new bit rate / original bit rate)

The video encoder 18 re-encodes the video data with a modified quantizer matrix (new Quant), allowing the bit rate to be changed without sacrificing the improvement in picture quality degradation. Generally, in an encoding operation prior to a decoding operation, the quantization step size of an inconspicuous portion of image data is increased to increase a compression ratio, while the quantization step size of a significant portion of image data is decreased to improve image quality. Modifying the quantizer matrix according to the above equation allows the bit rate to be changed while maintaining the above mentioned increase in compression ratio and improvement in image quality as it is.

Specifically, the inconspicuous portion of the image data encoded in the increased quantization step size also includes a lot of invisible noise, while the significant portion encoded in the reduced quantization step size contains many noise-free signal components. There is no need to re-encode noise in insignificant parts at all, thus allowing the quantization step size to be increased to improve the compression ratio. Conversely, the apparent part needs to minimize the growth of noise, so the quantization step size needs to be reduced. Therefore, changing the quantizer matrix (Quant) with the above equation enables re-encoding in the encoder 18 without sacrificing image quality.

The method of avoiding image quality degradation with the same parameters as those used in the previous encoding operation can also be used in conventional systems with two independent units being an MPEG stream reproducer with a monitor and for monitoring the reproduction Editor for image recoding. This type of system passes the video and audio signals reproduced by the reproducer to the editor. In the renderer, the video signal passed to the editor is changed by switching operations.

When using the above method, a parameter transfer cable is connected between the reproducer and the editor for transferring five parameters, or three parameters representing information about the image type, quantizer matrix (Quant), and DCT type. In order to prevent the relationship between parameters and video frames from being ambiguous, the ID and PTS contained in the reproduced stream are sent to the editor, and also add the ID and PTS to the parameters sent over the parameter transfer cable.

Fig. 7 shows an editing system capable of re-recording audio data according to a fourth embodiment.

The editing system of this embodiment includes a demultiplexer 25 for dividing stream 1 into video packets and audio packets; a demultiplexer 29 for dividing stream 0 into video packets and audio packets; and a switch 27 and 28 for selecting one of the outputs from video encoder 18, demultiplexer 25 and demultiplexer 29, from audio encoder 19, demultiplexer 25 and demultiplexer Select one of the outputs from device 29. The other structures are the same as those in the second embodiment, so a detailed description thereof is omitted here.

The outputs of switches 27 and 28 are compressed by multiplexer 26 and output to flow controller 12 in the form of a stream. Specifically, as shown in FIG. 8, segments A, B, and C of stream 3 are formed in a manner similar to that of the first embodiment and segment D of stream 3 is formed by selecting the segment of stream 1 by switch 27. Video packets in D, audio packets in segment D of stream 0 are selected by switch 28, resulting in a stream consisting of video packets in stream 1 and audio packets in stream 0. Like the first embodiment, it is not possible to rewrite the segment A of the stream 0 into the DVD-RAM 11.

The editing operation according to the fifth embodiment will be described below, which is performed when an editor edits a stream read out from a DVD-RAM while watching a monitor.

The editing operation of the first embodiment is performed when the editing point is known in advance. Therefore, segment B of stream 0 can be re-encoded according to the front of the segment. However, when an editor sets an edit point while observing a monitor, the data of the edit point is not input into the editing system in advance. Editing operations in this case are described below with reference to FIGS. 1 and 2 .

The editing operation for editing section A of stream 0 is the same as in the first embodiment. When segment B of stream 0 is edited, the editing system does not know the edit point, so that the editing system cannot recognize segment B of stream 0. Therefore, the same editing operation as in the processing of Section A is continued, thereby recording the sequence with respect to Sections A and B in Stream 0 in the DVD-RAM 11 .

When the editor determines the editing point, the timing controller 13 operates the switches 23 and 24 to connect the audio encoder 19 and the video encoder 18 to the AV decoder 21, and operates the switch 16 to connect the multiplexer 16 to the stream control Device 12 is connected. Therefore, in video encoder 18 and audio encoder 19, the data of segment C in stream 1 decoded by AV decoder 21 is re-encoded, and then passed through multiplexer 17, switch 16 and stream controller 12 , transmit and record it in DVD-RAM11. The editing operation for editing section D in stream 1 is the same as in the first embodiment.

After completing the decoding of segment C of stream 1, timing controller 13 operates switch 16 with the same method as the first embodiment, and the segment D of stream 1 output by delay circuit 15 is directly delivered to stream controller 12, thereby Section D is recorded in DVD-RAM 11 following Section C.

Once the edit point is determined, segment B of stream 0 can be ranged as described above. Therefore, after completing the editing of Stream 0 and Stream 1 by the above method, the editing system reads out the data in Section B of Stream 0 again, decodes, encodes and rewrites it to the corresponding area of DVD-RAM 11.

As described above, when the edit point is not predetermined, stream 0 and stream 1 are first rewritten in the DVD-RAM 11 without re-encoding of segment B of stream 0, and then segment B of stream 0 is re-encoded so that the stream interpolation. Typically, segment B is 0.5 seconds in length. Therefore, the editor feels that the editing operation is performed in real time.

If segment B of stream 0 is not re-encoded, the reproduced image will be distorted. This is because, as discussed in Fig. 10(a) and Fig. 10(b), the sequence of encoded pictures is different from the frame sequence of original pictures, so when cutting a GOP, it will cause several Frames disappear. The re-encoding of segment B of stream 0 avoids image distortion during reproduction.

Although the present invention has been disclosed in terms of preferred embodiments for ease of understanding, it should be understood that the invention can be implemented in various ways without departing from the principles of the invention. Accordingly, the invention should be understood to include all possible embodiments as well as modifications to the illustrated embodiments, which can be practiced without departing from the principles of the invention as defined in the appended claims.

Claims (17)

1. A stream editing device, characterized in that the device is designed to switch one of two input streams used in editing operations to the other at an edit point set in each stream, said device include: a first decoder, configured to decode the first stream; a second decoder for decoding the second stream; an encoder for re-encoding at least one of the first and second streams decoded by said first and second decoders; a controller for controlling editing of the first and second streams so as to produce a third stream consisting of the pre-segment preceding the edit point in the first stream and the tail segment following the edit point in the second stream When generating the third stream, the controller combines a part of at least one of the preceding segment and the trailing segment in the first and second streams with the first and second streams Merging other parts of said front segment and said tail segment, wherein said part of said at least one segment is decoded by a corresponding one of said first and second decoders and by said encoder re-encoding, said portion being defined as having a given length away from an edit point set in one of the first and second streams, said other portion referring to the segmented portion before decoding and re-encoding; a header generator for generating a series of headers for the third stream; demultiplexing means for dividing the first and second streams into video packets and audio packets; and Multiplexing means for multiplexing video packets in one of the first and second streams and audio packets in the other stream. 2. The stream editing device of claim 1, wherein both the first and second streams are composed of a plurality of groups of pictures, and the controller determines the The length of the portion in that starts at the edit point and ends at the front of the group of images the edit point is in. 3. The stream editing device according to claim 1, wherein the encoder uses encoding information obtained by decoding at least one of the first and second streams by the first and second decoders to encode the first and at least one of the second streams is re-encoded. 4. The stream editing device according to claim 3, wherein the encoding information includes a picture type, a quantizer matrix, and a discrete cosine transform type. 5. The stream editing device according to claim 4, wherein when the encoder is required to re-encode at least one of the first and second streams, change the When encoding a stream at a bitrate, the encoder uses a quantizer matrix that varies according to the following relationship: Quantizer matrix after change=Quantizer matrix before change×(Bit rate after change/Bit rate before change). 6. The stream editing device as claimed in claim 3, wherein the decoded video and/or audio data and encoding information are transferred from said first and The second decoder is passed to the encoder. 7. The stream editing device according to claim 1, wherein when the part to be incorporated into the third stream is defined in the tail section of the first stream, the encoder The part is re-encoded after the second-rate editing is finished. 8. The stream editing device according to claim 1, characterized in that, in the process of generating the third stream, the controller selects the following four parts, one of which is the front segment in the first stream before decoding and re-encoding The second part is the second part of the front segment of the first stream output by the encoder after the first part and before the edit point, and the third is the second part of the first stream output by the encoder after the edit point The first part of the trailer segment in the stream, which is the second part of the trailer segment in the second stream after the first part of the second stream and before decoding and re-encoding. 9. The stream editing device according to claim 8, wherein both the first and the second streams are composed of a plurality of image groups, and the controller determines the range of the second part of the first stream and the range of the second stream of the second stream. The range of the first part, where the second part of the first stream starts from the edit point to the front of the group of pictures in which the edit point is located, and the first part of the second stream starts from the edit point to the end of the group of pictures in which the edit point is located. 10. A stream editing method, which can switch one of the two input streams used in the editing operation to the other at an edit point set in each input stream, characterized in that the method comprises the following step: decoding the first and second streams with two decoders; re-encoding for at least one of the first and second streams decoded in said decoding step; and Editing of the first and second streams is controlled so as to produce a third stream consisting of a front segment located before the edit point in the first stream and a tail segment located after the edit point in the second stream, so that The control step combines a part of at least one of the preceding segment and the trailing segment in the first and second streams with the preceding segment in the first and second streams when generating the third stream merged with other parts of said trailing segment, wherein said part of said at least one segment is decoded and re-encoded, and is defined to have a segment away from an edit point set in one of the first and second streams for specified length, said other part refers to the fragmented part before decoding and re-encoding; generate a sequence of headers for the third stream; dividing the first and second streams into video packets and audio packets; and Video packets of one of the first and second streams and audio packets of the other stream are multiplexed. 11. A stream editing method as claimed in claim 10, characterized in that both the first and the second stream consist of a plurality of groups of pictures and are decoded and re-encoded and will be merged into the third stream The length is defined from the start of the edit point to the front of the image group in which the edit point is located. 12. The stream editing method according to claim 10, wherein the encoding step uses the encoding information obtained by decoding at least one of the first and second streams in the decoding step to encode the first and second streams. At least one of the streams was re-encoded. 13. The stream editing method according to claim 12, wherein the encoding information includes an image type, a quantizer matrix, and a discrete cosine transform type. 14. The stream editing method according to claim 13, wherein when said encoding step is required to change the bit rate used when encoding one of the first and second streams, said encoding step uses The quantizer matrix changed by the following relation: Quantizer matrix after change=Quantizer matrix before change×(Bit rate after change/Bit rate before change). 15. The stream editing method according to claim 10, wherein when the part to be merged into the third stream is defined in the trailing section of the first stream, the encoding step is performed on the second stream. The part is re-encoded after the second-rate editing is completed. 16. The stream editing method according to claim 10, characterized in that, in the process of generating the third stream, the control step selects the following four parts, one of which is the front segment in the first stream before decoding and re-encoding , the second is the second part of the pre-segment after the first part and before the edit point, decoded and re-encoded in the first stream, and the third is the tail segment after the edit point, in the decoded and re-encoded second stream The first part of the segment, which is the second part of the trailing segment in the second stream after the first part of the second stream and before decoding and re-encoding. 17. The stream editing method according to claim 16, wherein both the first and second streams are composed of a plurality of image groups, and the range of the second part of the first stream and the range of the first part of the second stream are determined, The second part of the first stream is from the edit point to the front of the image group where the edit point is located, and the first part of the second stream is from the edit point to the end of the image group where the edit point is located.

Images