KR20000032824A - Method for converting compressed moving pictures in an image system - Google Patents

Abstract

PURPOSE: A method for converting compressed moving pictures in an image system is provided so that a moving picture compression code of the H.263 is directly converted to that of MPEG-2 according to a compression algorithm common to different kinds of moving images in a variable-length code, a complete compression area of an image system, and converting a syntactic structure causing a failure in the conversion between two codes and parameters required for other processes. CONSTITUTION: With respect to a firstly inputted image of a first stream, a sequence hierarchy of a second stream and a hierarchy of a group of pictures are generated by analyzing header information. With respect to the other images of the first stream, a compressed moving picture of the first stream is converted to a compressed moving picture of the second stream in a variable-length code, a complete compression area, by generating a hierarchial information and changing a certain particular parameter values necessary for each hierarchy.

Description

Compressed Video Conversion Method in Image System

According to the present invention, the H.263 standard video compression code is converted into a standard video compression code conforming to the form of MPEG-2 (MPEG-2) in the variable length code, which is the fully compressed region of the video system. The present invention relates to a compressed video conversion method that enables easy conversion between videos.

In general, the conversion between heterogeneous image codes may be defined as a function that enables mutual compatibility by converting a code system made using different compression schemes into a desired compression syntax. This is called a transcoder, and the most common transcoder is a method of a pixel region that decodes a compressed code into a linear pulse code modulation (PCM) signal and encodes the compressed code into a compressed code that needs to be interlocked again. The processing method in such a pixel region is shown in FIG.

1 is a diagram illustrating a code conversion process in a conventional pixel area.

The transcoder of the pixel region is the most easily accessible method of converting heterogeneous images, but this requires a decoder and encoder pair for conversion, which is a dual economic burden regardless of whether a function exists in a network node or a user terminal. In addition to the loss and complexity of implementation, the transcoder of the pixel region may be used, and thus, the deteriorated image quality of the converted image may occur, and in particular, an unacceptable delay may occur in multimedia communication.

As a conventional method, there is a method (GIF to JPEG) for converting a Graphics Interchange Format (GIF) to a Joint Photographics Expert Group (JPEG).

Here, GIF is an image and graphic standard for general computers. It converts high-resolution information according to its purpose using Palette, a look-up table, and converts color information during this process. The index value to be expressed is compressed by the "Lempel-Ziv-Welch (LZW) algorithm".

On the other hand, JPEG is a standard for compressing continuous-tone still images such as photographs. This algorithm aims to convert GIF, which is widely used for displaying image and graphic information, to JPEG on the World Wide Web (WWW) of the Internet, and to reduce the file size by 10-50%. .

On the other hand, the research on the "Motion JPEG to MPEG-1" conversion has changed the compression code into a form that an MPEG-1 decoder can process an MJPEG video composed of independent Intra frames. For example, it does not generate the I, P (Predictive coded picture), B (Bidirectional predictive coded picture) frame, or motion vector of the MPEG-1, but changes in syntax such as a sequence and a group of picture header. Simple structure for changing some parameters. The main focus of this research is to enable the video editing function in MPEG-1 as in JPEG, rather than converting to a complete MPEG-1 compression code.

Besides, there is a study of converting MPEG-1 / 2 video into MJPEG in the compressed region, but this is also a method in which image editing is a main application.

As such, the video compression method uses H.261 of the "Telecommunication standardization sector of International Telecommunication Union" (ITU-T) for interactive communication and "ISO / IEC (International Organization Standardization / International) for storage and distributed communication. Electrotechnical Commission ("Electrotechnical Commission") has clearly identified applications for MPEG-1, but H.263 and MPEG-2, which have been standardized for further study, have been extended to cover all multimedia communication services. This means that two compression methods can be used for the same purpose, and the compressed video algorithms of these two international standards organizations (ITU-T, ISO / IEC) are dependent on the performance of the compression method in interdependent relationships that complement each other's functions. The superiority was thus transformed into the competition of choice.

H.263 and MPEG-2, which are recommended by the International Organization for Standardization, are compression methods that target the same video, but cannot interoperate. That is, H.263 encoded streams cannot be processed by the MPEG-2 decoder, and MPEG-2 streams cannot be decoded by H.263. This did not cause major problems in H.261 and MPEG-1, which were primarily used for video telephony, video conferencing, and storage, but extended to real-time interactive communications, broadcast, communication services, and storage communications. H.263 and MPEG-2 inevitably use these compression schemes for the same purpose, and the inability to interoperate between them causes economic loss on the user side and heterogeneity on the application side.

This heterogeneity of the application side can be solved by using a corresponding decoder and a transcoder in a general pixel area connecting the encoder in multiple stages, but this also can be duplicated regardless of the position of the transcoder (network or user terminal). It is an economic burden. In addition, the use of a transcoder, which is a multi-stage codec, causes deterioration in image quality, particularly excessive processing delay, thereby limiting real-time communication.

Unlike H.261 and MPEG-1, which failed to spread the generalized communication service due to network construction and immaturity of the multimedia market, Public Switched Telephone Network (PSTN) and Local Area Network (LAN) H.263 and MPEG, considering the current situation where the network infrastructure, such as ATM (ATM), Asynchronous Trasfer Mode (ATM), Cable Television (CATV) and digital airwaves is mostly completed, and the public's familiarity with multimedia, Various multimedia communication services using -2 will spread rapidly.

Accordingly, mutual interworking between these heterogeneous compression codes, which is currently impossible, has become a problem that must be solved.

Resolving heterogeneity from the perspective of application services and interworking of heterogeneous compression codes to provide economic advantages must be provided to ensure the efficiency that can be realized at an affordable price and the real time processing required for multimedia communication.

As mentioned above, unlike the existing standard, which clearly distinguished and recommended the field of application of video compression code, it is mainly used now and the compression code algorithm newly researched is targeted for the same application. That is, if these heterogeneous video compression schemes are used for the same multimedia communication, application services between endpoints cannot be interworked.

In an attempt to solve the problem of heterogeneity due to the difference of video compression methods, there have been studies of ITU-T and ISO / IEC, but this is also limited to unidirectional interworking that accepts the basic compression code of H.263 in MPEG-4. So far, the interworking from MPEG-4 to H.263 has not been resolved.

In particular, since H.263 and MPEG-2, which are representative video compression algorithms, cannot be interoperable, these heterogeneous compression codes cannot be used when they are used in multimedia application services including interactive, communication, and broadcasting. In addition to the problem, in order to receive a specific video application service, there is a problem of economic loss of the user side that requires a separate video compression processing unit for each service.

In order to solve the above problems, the present invention utilizes the information generated according to a compression algorithm common among heterogeneous videos in a variable length code, which is a completely compressed region of an image system, and converts two other codes. Compression video conversion method for converting H.263 standard video compression code into standard video compression code conforming to MPEG-2 form by converting the necessary syntax structure and other processing parameters, and a computer recording a program for realizing the same. The purpose is to provide a recording medium that can be read by.

1 is a diagram illustrating a code conversion process in a conventional pixel region.

2 is a flowchart illustrating an embodiment of a method for converting a compressed video according to the present invention.

3 is an explanatory diagram showing the conversion relationship for each layer of an H.263 stream and an MPEG-2 stream used in the present invention;

4 is an explanatory diagram showing a conversion relationship between a macroblock layer of an H.263 stream and an MPEG-2 stream used in the present invention;

5 is an explanatory diagram showing a motion vector encoding process of an H.263 stream used in the present invention.

6 is an explanatory diagram showing a motion vector encoding process of an MPEG-2 stream used in the present invention.

7 is an explanatory diagram of a DC coefficient and a negative vector initialization process of an MPEG-2 stream used in the present invention.

8 is an explanatory diagram of a DC encoding process for a block layer in an MPEG-2 stream used in the present invention.

The present invention for achieving the above object, in the compressed video conversion method applied to the video system, for the first input video of the first stream, by analyzing the header information of the sequence layer and video group (GOP) of the second stream A first step of creating a layer; And compressing the compressed video of the first stream in the variable length code, which is a completely compressed region, by generating hierarchical information on the remaining images of the first stream and changing specific parameter values required for each layer. The second step of converting to a video.

In addition, the present invention, in the video system having a processor, the video system having a processor, the header information for the first input video of the first stream, the header information and the video group (GOP) of the second stream Creating a layer; And compressing the compressed video of the first stream in the variable length code, which is a completely compressed region, by generating hierarchical information on the remaining images of the first stream and changing specific parameter values required for each layer. Provided is a computer-readable recording medium having recorded thereon a program for realizing a function of converting a video.

Hereinafter, with reference to the accompanying Figures 2 to 8 will be described in detail a preferred embodiment according to the present invention.

2 is a flowchart illustrating a method of converting a compressed video according to the present invention.

As shown in FIG. 2, the compressed video conversion method for converting an H.263 compressed video into an MPEG-2 compressed video according to the present invention first analyzes the header information of the H.263 stream (201), and then determines the first video ( A sequence layer and a group of picture (GOP) layer of an MPEG-2 stream are generated for a picture (203).

Thereafter, for the remaining images of the stream (208), frames of the H.263 stream and the MPEG-2 stream are matched to generate an MPEG-2 video layer (204).

Next, the MPEG-2 slice layer is generated by analyzing the resolution of the H.263 stream and the structure of the macroblock (205).

Thereafter, the information of the macroblock in the H.263 stream is analyzed and converted into an MPEG-2 macroblock layer (206).

Finally, the block layer information of the H.263 stream is analyzed and converted into the block layer of the MPEG-2 stream (207).

The compressed video conversion method for converting H.263 compressed video into MPEG-2 compressed video according to the present invention is a new method that satisfies the requirements of a transcoder. Based on the motion estimation, compensation, Discrete Cosine Transform (DCT) and entropy coding according to Pulse Code Modulation (DPCM), information generated according to a common compression algorithm Is to make the most of it, and converts the syntax and other parameters necessary for processing other obstacles.

Accordingly, the present invention converts the compressed code of H.263 into a video code suitable for MPEG-2 format, and performs the processing on the fully compressed syntax, which has the advantages of image quality improvement and real-time conversion, and is particularly easy with simple processing. It is an efficient way to implement it.

A compressed video conversion method for converting an H.263 compressed video according to the present invention as described above into an MPEG-2 compressed video will be described in more detail with reference to FIGS. 3 to 8.

3 is an explanatory diagram showing the conversion relationship for each layer of an H.263 stream and an MPEG-2 stream used in the present invention.

The present embodiment can convert between heterogeneous video codes by a simple procedure of generating necessary layer information on a fully compressed code and changing specific parameter values required for each layer.

As shown in FIG. 3, unlike the H.263 stream 3a, the MPEG-2 stream 3b has a hierarchical structure such as the sequence layer 301 and the GOP layer 302 for image manipulation and tolerance of error information. Has Therefore, in order to convert the compressed code, some header information and frame structure that do not exist in H.263 must be generated.

The sequence layer 301 is the highest layer of the compressed video of the MPEG-2 stream 3b. The sequence layer 301 provides information about a resolution, a frame structure, an output bit rate, an output buffer buffer (VBV) parameter, and quantization processing. Since the header information and the main parameter values included in the sequence layer 301 are generated according to the result of analyzing the input image of the H.263 stream 3a. Since the quantization matrix not included in the H.263 stream 3a is an important factor in determining the image quality of the stream converted into the MPEG-2 stream 3b, the linear quantization characteristic of the H.263 stream 3a is preserved as it is. The matrix value is converted into a matrix value and written in the header of the changed MPEG-2 stream 3b to process the DCT coefficient of the H.263 stream 3a without loss.

The part to be processed in the video layer 303 is a frame matching process of the H.263 stream 3a and the MPEG-2 stream 3b. The contents of the video header are changed to correspond to P and B. In addition, a temporary reference value (TR) is converted into a serial number having a period of 1024 in the order of MPEG-2 frames, and an f code indicating a motion search range belonging to the video layer 303 is converted into an H.263 stream 3a. When the processed stream is set according to the information of the decoding, it is displayed normally on the screen through accurate motion compensation. In addition, the parameters of the video header are set in a linear quantization scale mode that can deliver the quantization characteristics of the H.263 stream 3a without loss.

In H.263 stream 3a, a fixed number of macroblock groups in a row is called a group of blocks (GOB), and this GOB layer is a slice layer 304 of MPEG-2 stream 3b. This is the layer corresponding to.

However, the handling of GOB in H.263 stream 3a is optional and this layer 304 may not exist.

Therefore, the GOB layer 302 cannot be directly converted into the slice layer 304 of the MPEG-2 stream 3b, and the number of pixels corresponding to the horizontal axis of the image is determined based on the resolution information of the H.263 stream 3a. The slice layer 304 is generated by counting the number of macroblocks corresponding to the value divided by 16.

The macroblock is used as a basic unit of motion estimation and compensation. Basically, the macroblocks of the H.263 stream 3a and the MPEG-2 stream 3b are similar, but the type and encoding method of the generated macroblocks are similar. Has a difference and requires a conversion process. That is, the conversion of the macroblock layer 305 is necessary to confirm the existence of the code of the macroblock for the H.263 stream (3a), and to analyze the code shape and the motion vector.

Accordingly, the conversion of the macroblock layer 305 is based on the results of the macroblock address increment (MBAI: MacroBlock Address Increment) value and the coded block pattern (CBP) of the MPEG-2 stream 3b. Create and process motion vectors.

The conversion of the macroblock layer 305 is performed separately for intra and inter blocks having motion information, and the intra macroblock of the H.263 stream 3a that does not include dual motion information is directly converted to an MPEG-2 stream ( It is changed to the intra macroblock of 3b). The interblock divides the macroblock having only forward motion and the macroblock including bidirectional motion information according to the motion information determined according to the frame type of the PB, thereby forming a PB frame of the MPEG-2 stream 3b. Changed to correspond to the block. The conversion relationship of the macroblock layer 305 between the H.263 stream 3a and the MPEG-2 stream 3b is shown in FIG.

4 is an explanatory diagram showing a conversion relationship between a macroblock layer of an H.263 stream and an MPEG-2 stream used in the present invention.

The types of macroblocks in H.263 stream (3a) are coded macroblock indication (COD), macroblock type & coded block pattern for chrominance (MCBPC), coded block pattern for luminance (CBPY), and motion vector data (MVD). While they have unique meanings depending on their existence and configuration, macroblocks in the MPEG-2 stream 3b are precisely classified by type.

Therefore, a process of analyzing the information included in the macroblock of the H.263 stream 3a and matching it to the form of the macroblock corresponding to the MPEG-2 stream 3b is necessary.

Referring to FIG. 4, in the H.263 stream 3a, "COD" indicates whether a corresponding macro block exists or not, and "MCBPC" distinguishes between intra and inter block types and exists in the presence or absence of a code of a color difference signal and a macro block unit. It includes a quantization parameter. &Quot; CBPY " indicates the presence or absence of a code of the luminance signal. Of these, "MCBPC" and "CBPY" are used to generate CBP information of the MPEG-2 stream 3b.

The macroblock included in the intra frame of the first coded H.263 stream 3a is changed to be the same as the macroblock constituting the I frame of the MPEG-2 stream 3b. In the case of other PB frames, the macroblock is subdivided into intra and inter, and it is again determined whether each has a quantization parameter, thereby changing to a proper macroblock type of the MPEG-2 stream 3b. In the case of the dual inter macroblock, the CBP of the MPEG-2 stream 3b is generated by using the MCBPC and CBPY of the H.263 stream 3a, and the motion information is added to the corresponding macroblock of the H.263 stream 3a. Is determined, and it is changed to one of three macroblock types in which CBP, a motion vector, and a CBP and a motion vector are present.

The H.263 stream 3a and the MPEG-2 stream 3b use similar algorithms for motion estimation and compensation, but the method of encoding the resulting motion vector is different. The motion vector encoding process of the H.263 stream 3a and the motion vector encoding process of the MPEG-2 stream 3b are illustrated in FIGS. 5 and 6, respectively.

5 is an explanatory diagram illustrating a motion vector encoding process of an H.263 stream used in the present invention, and FIG. 6 is an explanatory diagram illustrating a motion vector encoding process of an MPEG-2 stream used in the present invention. "Current Motion Vector", "MV1" is Previous Motion Vector, "MV2" is Upper Motion Vector, and "MV3" is Upper Right Motion Vector Vector). The dotted line represents the image or GOB border.

As shown in Figs. 5 and 6, the encoding of the motion vector of the H.263 stream 3a corresponds to the median value of the actual motion vector value selected by the motion estimation and the vector of the neighboring macroblock. Expressed as a difference from. At this time, the vector used for the intermediate value should be the actual value extracted during the search.

On the other hand, the encoding of the motion vector in the MPEG-2 stream 3b is represented by a differential value from the motion vector of the previous macroblock. That is, in the MPEG-2 stream 3b, the range of the motion vector is determined by the f code, which is represented by the difference value of the motion vector and the remaining value represented by f code-1.

Therefore, in order to convert the correct motion vector, after calculating the original correct motion estimation value from the motion vector included in the macroblock of the H.263 stream 3a, the motion vector is used by the method used in the MPEG-2 stream 3b. Must be recoded.

Note that when converting motion vectors, macroblocks must exist at the beginning and end of slice layer 304 in MPEG-2 stream 3b. After analyzing the shape of the macroblock of the input H.263 stream 3a for this purpose, if the first and last macroblocks are omitted, only the motion vector generates a macroblock having a value of 0, and the rest of the macroblocks are the previous macroblocks. Expressed as and difference.

As described above, the CBP constituting the macroblock is generated by using the MCBPC and CBPY of the H.263 stream 3a and using a lookup table that distinguishes intra and inter blocks.

The conversion of the macroblocks must be accompanied by an initialization process that matches the characteristics of the MPEG-2 stream 3b. This initialization process is applied to the encoding of the DC coefficient and the motion vector represented by the difference with respect to the previous macroblock, and under the conditions as shown in FIG. 7, the value of the previous macroblock is set to 0, and the difference is original. DC coefficients and motion vector values of the macroblock

8 is an explanatory diagram for a DC encoding process of an intra block of an MPEG-2 stream used in the present invention.

The block layer 306 conversion from the H.263 stream 3a to the MPEG-2 stream 3b converts the DC coefficient of the intra block and the DC (Direct Current) and AC (AC) alternative coefficients of the inter block. Process separately.

The DC coefficients of the intra blocks in the H.263 stream 3a are sequential representations with 8-bit fixed-length code values, while the MPEG-2 streams 3b are represented with differential values with variable-length codes, with luminance and chrominance components. It is coded differently according to.

Therefore, the DC of the intra block of the H.263 stream 3a is divided according to the luminance and the chrominance component, and then expressed as a difference value, and then converted into the VLC of the MPEG-2 stream 3b. Here, Variable Length Code (VLC) and Run-Length Code (RLC) coding to be used are the same as the processing of the interblock to be described later.

The RLC process for the DC and AC coefficients of the interblock in the H.263 stream 3a is as follows.

In general, DCT coefficients are represented by successive zero and nonzero coefficient values, each referred to as RUN, LEVEL. However, the RLC encoding representing this uses a method in which the H.263 stream 3a and the MPEG-2 stream 3b are different from each other. That is, the MPEG-2 stream 3b performs a two-dimensional VLC indicating the end of the code by using the end of block (EOB), while the H.263 stream 3a performs the last (in addition to RUN and LEVEL). LAST) using a three-dimensional VLC.

Therefore, the EOB information is added to the DC and AC coefficients included in the interblock of the H.263 stream 3a when the information of the LAST portion is checked and converted into the MPEG-2 stream 3b.

As described above, the present invention enables direct conversion of the compression code of the H.263 stream 3a, which is currently a representative compression code, into a video code that conforms to the form of the MPEG-2 stream 3b, thereby enabling video conferencing or on-demand. When using these compression codes for VOD (Video On Demand), etc., the heterogeneous code is much simpler than the existing method where the user does not receive the service or requires a complicated compression code processing device to receive the service. It is possible to convert between them, and to improve image quality, in particular, real-time processing, which is a requirement for multimedia communication.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains, and the above-described embodiments and accompanying It is not limited to the drawing.

As described above, the present invention does not perform the motion and discrete cosine transform (DCT) related processes in converting the H.263 video compression code into the MPEG-2 compression code. Real-time processing, which is an essential requirement for image quality improvement and multimedia communication, can be implemented, and various multimedia communication services can be provided without using a separate device when using these heterogeneous compression codes for video conferencing or video on demand. It can be effective.

Claims (10)

In the compressed video conversion method applied to the imaging system, A first step of generating a sequence layer and a video group (GOP) layer of the second stream by interpreting header information with respect to the first input image of the first stream; And For the remaining images of the first stream, hierarchical information is generated, and specific parameter values required for each layer are changed to convert the compressed video of the first stream into the compressed video of the second stream in a variable length code that is a completely compressed region. Second step to convert Compressed video conversion method comprising a. The method of claim 1, The first stream, Compressed video converting method, characterized in that the H.263 stream. The method of claim 1, The second stream, MPEG-2 (MPEG-2: Moving Picture Expert Group-2) A compressed video conversion method characterized in that the stream. The method according to any one of claims 1 to 3, The sequence layer of the first step, The highest layer of the compressed video of the MPEG-2 stream, which includes information about the resolution of the image, the structure of the frame, the output bit rate, the output buffer buffering (VBV) parameters, and the quantization process. Compressed video converting method, characterized in that the generated header information and the main parameter value according to the result of analyzing the input video of the H.263 stream. The method of claim 4, wherein The second step, A third step of generating the MPEG-2 image layer by matching frames of the H.263 stream and the MPEG-2 stream with respect to the remaining images of the H.263 stream; Analyzing the resolution of the H.263 stream and the structure of the macroblock to generate the MPEG-2 slice layer; A fifth step of analyzing the macroblock layer in the H.263 stream into a macroblock layer of the MPEG-2 stream by analyzing information of the macroblock; And By analyzing the block layer information in the H.263 stream and converting the block layer information into the block layer of the MPEG-2 stream, the compressed video of the H.263 stream is converted into the MPEG-2 stream in a variable length code that is a completely compressed region. 6th step of converting your video to compressed video Compressed video conversion method comprising a. The method of claim 5, The fourth step, And counting the number of macroblocks corresponding to a value obtained by dividing the number of pixels corresponding to the horizontal axis of the image by 16 based on the resolution information of the H.263 stream, and generating the slice layer. The method of claim 5, The fifth step, For the macroblock layer of the H.263 stream, a macroblock address increment (MBAI) value and a coded block pattern (CBP) are generated by checking whether a code exists in a macroblock and interpreting a code form and a motion vector. And a motion vector is processed to convert the MPEG-2 macroblock layer. The method of claim 7, wherein The motion vector is It is expressed by the difference between the actual motion vector value selected by the motion estimation at the time of encoding the H.263 stream and the median value of the vector of the neighboring macroblock, and the previous macro at the time of encoding the MPEG-2 stream. Compressed video conversion method characterized in that expressed as a differential value (differential value) with the motion vector of the block. The method of claim 5, Converting the block layer of the MPEG-2 stream of the sixth step, The DC (Direct Current) of the intra block of the H.263 stream is classified according to the luminance and the chrominance component, and then expressed as a differential value. Then, the variable length code (VLC) of the MPEG-2 stream is expressed. The compressed video conversion method characterized in that the conversion. In an imaging system with a processor, A function of generating a sequence layer and a video group (GOP) layer of the second stream by interpreting header information with respect to the first input video of the first stream; And For the remaining images of the first stream, hierarchical information is generated, and specific parameter values required for each layer are changed to convert the compressed video of the first stream into the compressed video of the second stream in a variable length code that is a completely compressed region. To convert A computer-readable recording medium having recorded thereon a program for realizing this.