KR100310774B1 - Image encoding apparatus - Google Patents

Abstract

The present invention provides an image signal including texture information and shape information for each processing macroblock which may include M * M (M is positive) pixels and may be divided into P (P is positive) DCT blocks of the same size. An image encoding apparatus for encoding, comprising: shape encoding means for generating coded shape information by encoding shape information for a processing macroblock, and decoding the coded shape information to generate reproduction shape information; The block type of the processing macroblock is determined by using the reproduced shape information from the shape coding means, and a block designation signal indicating the block type is provided. Shape checking means composed of a block and a boundary block including at least one object pixel and a background pixel; If the processing macroblock is determined as a boundary block based on the block designation signal from the shape checking means and the texture information in the video signal, the DCT type of the processing macroblock is determined by a forward coding method and the determined DCT type information is provided. DCT type determining means for determining the DCT type of the processing macroblock based on texture information in the video signal when it is determined as the boundary block or the object block and providing the determined DCT type information; And texture encoding means for encoding the texture information by a progressive encoding technique or a progressive encoding technique in response to the DCT type information from the DCT type determination means.

Description

Image coding device {IMAGE ENCODING APPARATUS}

The present invention relates to an object-based encoding system for dividing an image signal into shape information and texture information and then compressing and coding at low data rates. In particular, the present invention relates to an image encoding apparatus for determining a DCT type for texture information using shape information. It is about.

In general, in a digital video system such as video telephony and video conferencing, since a video signal is composed of a series of digital data called pixel values, a large amount of digital data is required to define each video signal. However, since the usable frequency band in a typical transmission channel is limited, in order to transmit a large amount of digital data through such a transmission channel, it is necessary to reduce the transmission amount of data using various compression techniques. In particular, in the case of low-rate video signal encoders such as video telephony and video conferencing systems, the need for data compression is more urgent.

One of the coding methods for encoding video signals in a low transmission coding system is the so-called object-oriented analysis-systhesis coding technique, in which the input video image is a plurality of objects. The three sets of parameters for defining the motion, contour, and pixel data of each object are processed through different coding channels.

An example of such an object-oriented coding scheme is the so-called Motion Picture Experts Group (MPEG-4), which is an audio and visual coding standard that allows for content-based interactivity. And improve coding efficiency and / or overall access in applications such as low bit rate communications, interactive multimedia (e.g. games, interactive TV, etc.), and area monitoring devices. See, eg, MPEG-4 Video Verification Model Version 2.0, International Organization for Standardization, ISO / IEC / JCT1 / SC29 / WG11 N1260, March 1996).

According to MPEG-4, a frame of an input video image is divided into a plurality of video object planes (VOPs) representing units that a user can access or manipulate in a bit stream. The VOP is represented as an object, and its width and height can be represented by a range of rectangles that are the minimum multiples of the 16 pixels (macroblock size) that surround each object, so that the encoder converts the input image image into VOP units. Will be processed.

The VOP disclosed in MPEG-4 includes shape information and texture information for an object corresponding to a plurality of macroblocks on the VOP, each macroblock having, for example, 16 * 16 pixels. Each macro block on the VOP may be classified into one of a background block, a boundary block, and an object block. The background block is in the VOP, but includes only background pixels that exist outside of the object, the boundary block includes one or more background pixels and one or more object pixels that exist inside of the object, and the object block includes only the object pixels. Shape information corresponding to a macroblock is encoded in macroblock units using a context-based arithmetic encoding (CAE) technique, for example, according to MPEG-4, and the texture information is DCT (Discrete Cosine Transform). ) Is encoded in units of macroblocks using existing coding methods such as quantization and variable length encoding. In particular, the DCT process for the texture information is performed in units of DCT blocks when the macro block is divided into four DCT blocks including 8 * 8 pixels.

As a result of the DCT process, one DC component and a plurality of AC components are generated for each DCT block. However, if all values of the texture information for the DCT block are constant, there is no AC component corresponding to the DCT block. Therefore, Coded Block Pattern Type (CBPY) information has been proposed to indicate whether the DCT block has at least one AC component. That is, if there is at least one AC component corresponding to the DCT block, the CBPY information has a bit of "1", otherwise it has a bit of "0". Thus, by detecting the CBPY information transmitted over the transport channel, it is possible to predict the presence of an AC component for the corresponding DCT block without further information about the corresponding DCT block and before the texture information for the corresponding DCT block is transmitted. have.

Meanwhile, in order to encode texture information about a VOP, existing encoding methods have first processed texture information for each macroblock using only a progressive encoding technique. However, when it is known that when the spatial and / or temporal correlation of a frame is lower than the spatial / temporal correlation of a field, it is known that the perturbed coding technique can improve the coding efficiency, thereby encoding the encoding information adaptively. The technique was introduced. Therefore, DCT type information indicating what encoding scheme was used to encode texture information was introduced to indicate the encoding state of the texture information, and the DCT type for each macro block is determined based on the texture information. For example, the image signal encoder determines the appropriate DCT type for the macro block by comparing the correlation between adjacent pixels in the macro block rearranged into the progressive block or the parallel block according to the progressive encoding technique or the progressive encoding technique. Thus, if it is determined that the forward coding scheme is more effective, the DCT type information for the macro block has a bit of "0", otherwise a bit of "1" is allocated to the DCT type information.

If the DCT type for the macro block is determined through a known method based on the texture information, the CBPY information for the macro block is obtained from the DCT result obtained by encoding the texture information for the macro block according to the determined DCT type. More specifically, if the macro block is a background block, the texture information is not encoded, and thus the DCT type information and the CBPY information are not generated. If the macro block is an object block, a forward coding technique or a progressive coding technique may be selected based on the texture information of the macro block to encode the texture information, and the macro block of the object block includes four DCT blocks to be encoded. Therefore, the CBPY information has 4-bit data in which each bit exclusively corresponds to one of four DCT blocks in the macro block. In other words, the object block contains only background pixels and does not contain any transparent blocks that need not be encoded. Meanwhile, if the macro block is a boundary block, the macro block may include a transparent block and a non-transparent block including at least one object pixel and to be encoded. Thus, the CBPY information corresponding to the boundary block may have i-bit data, where I is a positive number ranging from 1 to 4, and each bit is generated to correspond to each non-transparent block in the macro block.

1A to 1C show various examples of boundary macroblocks classified into two different shape blocks, namely, a progressive macroblock and a progressive macroblock. In the figure, the macroblocks P1 to P3 representing the progressive blocks are rearranged into a parallel block including the upper field blocks IT1 to IT3 and the lower field blocks IB1 to IB3. Therefore, in the progressive coding technique, the macro block is encoded based on the progressive macro block, and according to the progressive encoding technique, the macro block is encoded based on the parallel macro block including the upper and lower field DCT blocks.

As can be seen in FIG. 1A, since the progressive block P1 and the progressive block I1 corresponding to the progressive block P1 include only non-transparent blocks, their CBPY information is 4 bits regardless of the DCT type of the macroblock. Have data However, in Figs. 1B and 1C, the number of non-transparent blocks in a progressive block and a parallel block differs depending on their DCT type. Therefore, the number of bits of their CBPY information also changes due to the DCT type. That is, 2-bit CBPY information is generated when the macro block P2 is encoded through the progressive encoding technique, and 4-bit CBPY information is generated otherwise. On the other hand, when the macro block P3 is encoded by the progressive encoding technique, 2-bit CBPY information is generated, otherwise 1-bit CBPY information is generated.

In consideration of the above, if the macro block to be processed is a boundary block, the number of bits of CBPY information cannot be determined without knowing its DCT type.

However, in the existing encoding method including CBPY information and DCT type information, the data stream to be transmitted to the decoding side has a sequence as shown in FIG. 2. That is, the decoded shape information is first transmitted to the decoding side, and the other coded information follows shape coded in the order of CBPY information, DCT type information, and texture information. Therefore, when the corresponding macroblock is a boundary block and the data encoded in the above sequence is transmitted to the decoding side, the DCT type is determined based solely on the texture information, and the CBPY information depends on the DCT type information. There was a problem that the information cannot be reproduced correctly.

Accordingly, the present invention determines the DCT type for the texture information of the video signal by using the shape information as well as the texture information of the video signal, thereby predicting the DCT type through the shape information on the decoding side, thereby reproducing the CBPY information without error. It is an object of the present invention to provide an apparatus for encoding an image.

The present invention for achieving the above object, the texture information and shape for each processing macro block that includes M * M (M is a positive number) pixel, and can be divided into P (P is a positive number) DCT blocks of the same size An image encoding apparatus for encoding a video signal including information, comprising: shape encoding means for generating encoded shape information by encoding shape information for a processing macroblock, and decoding the encoded shape information to generate reproduction shape information; The block type of the processing macroblock is determined by using the reproduced shape information from the shape coding means, and a block designation signal indicating the block type is provided. Shape checking means composed of a block and a boundary block including at least one object pixel and a background pixel; If the processing macroblock is determined as a boundary block based on the block designation signal from the shape checking means and the texture information in the video signal, the DCT type of the processing macroblock is determined by a forward coding method and the determined DCT type information is provided. DCT type determination means for determining the DCT type of the processing macroblock based on texture information in the video signal when it is determined as the boundary block or the object block and providing the determined DCT type information; And texture encoding means for encoding the texture information by a progressive encoding technique or a progressive encoding technique in response to the DCT type information from the DCT type determination means.

1A to 1C show typical examples of boundary blocks classified into two different forms,

2 shows a sequence of typical data streams to be transmitted to a decoding side;

3 is an overall configuration diagram of an image encoding apparatus according to an embodiment of the present invention;

4 is a detailed configuration diagram of a DCT processor in the image encoding device of FIG. 3.

<Description of the code | symbol about the principal part of drawing>

100: shape check unit 110: shape encoder

120: DCT type determination unit 130: DCT processing unit

140: AC component detection unit 150: CBPY generation unit

160: CBPY encoder 170: VLC table group

180: texture encoder 190: multiplexer

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The image signal includes shape information and texture information of a plurality of macroblocks, and according to an embodiment of the present invention, the texture information is adaptively encoded in a macroblock unit through a forward coding technique or a progressive encoding technique. Each macro block contains M * M, for example 16 * 16 pixels, and can be divided into 4DCT blocks with the same number of pixels, for example 8 * 8 pixels. According to the present invention, an appropriate coding scheme for each macroblock, i.e., a progressive or a parallel encoding scheme, is determined based on texture information and shape information for the macroblock.

3 shows an image encoding apparatus according to an embodiment of the present invention.

In FIG. 3, texture information of the processing macroblock is input to the DCT type determination unit 120 and the DCT processing unit 130, respectively, and the corresponding shape information is applied to the shape encoder 110.

The shape encoder 110 encodes the shape information using the known shape coding method disclosed in MPEG-4, and then outputs the coded shape information to the multiplexer 190 through the line L12. In addition, the shape encoder 110 provides the shape check unit 100 and the DCT processing unit 130 through the line L10, respectively, with the reproduced shape information of the processing macro block generated by decoding the shape information.

The shape check unit 100 determines the block type for the processing macro block by using the reproduced shape information. The block type includes a background block, an object block, and a boundary block as described above. The background block includes only the background pixel, the object block includes only the object pixel, and the boundary block includes both the background pixel and the object pixel. When the block type of the processing macro block is determined, the shape checking unit 100 provides a block designation signal indicating the block type of the processing macro block to the DCT type determination unit 120.

The DCT type determination unit 120 determines the DCT type of the processing macro block in response to the texture information of the processing macro block and the block designation signal generated from the shape checking unit 100. That is, as exemplarily illustrated in FIGS. 1A to 1C, when the processing macro block is determined as the boundary block according to the block designation signal, the DCT type determination unit 120 performs progressive encoding on the texture information about the processing macro block. The DCT processor 130 provides a forward encoding designation signal to encode using only a technique. On the other hand, if the processing macro block is determined to be an object block, the DCT type determination unit 120 determines the DCT type of the processing macro block using a known DCT type determination method based on texture information, and indicates the DCT type of the processing macro block. The type information is provided to the multiplexer 190 through the line L14 and output to the DCT processing unit 130 through the line L16.

On the other hand, when the processing macro block is set as the background block, the DCT type determination unit 120 transmits a DCT disable signal to the DCT processing unit 130 through the line L16 so that the DCT processing unit 130 performs the DCT for the background block. Prevents the process from running Therefore, the DCT processing unit 130 does not operate when the DCT disable signal is input.

When the processing macro block is a background block, as another embodiment of the present invention, the corresponding texture block may not be input to the image encoding apparatus. In this case, since the texture block corresponding to the background block is not input to the video encoding apparatus, the video encoding apparatus does not perform any action.

For ease of understanding, the process related to the background pixel will be omitted below.

When the forward encoding designation signal or the DCT type information is applied to the DCT processing unit 130, the DCT processing unit 130 executes the existing discrete cosine conversion process using the texture information and the reproduction shape information of the processing macro block, thereby processing the processing macro. Generate a set of DCT coefficients for a block in units of DCT blocks.

4 shows a detailed configuration diagram of the DCT processing unit 130.

In FIG. 4, the DCT processing unit 130 includes a first reordering stage 132, a second reordering stage 134, a selection stage 136, and a DCT stage 138.

Referring to FIG. 4, the first reordering stage 132 receives reproduction shape information from the shape encoder 110 in response to a forward encoding designation signal or DCT type information applied through the line L16 from the DCT type determination unit 120. By reconstructing, the rearranged shape information is generated and the rearranged shape information is provided to the selection stage 136.

The second reordering stage 134 generates reordered texture information by reconstructing texture information in response to a forward encoding designation signal or DCT type information applied from the DCT type determination unit 120 through the line L16, and realigns the texture information. Information is supplied to the selection stage 136.

The selection stage 136 detects the non-transparent block for the processing macro block using the rearranged shape information and then provides the rearranged texture information corresponding to the detected non-transparent block to the DCT stage 138, which is non-transparent. The block has a DCT block size and includes at least one object pixel.

The DCT stage 138 converts the rearranged texture information corresponding to each non-transparent block into a DCT coefficient set and transmits the rearranged texture information to the AC component detector 140 through the texture encoder 180 and the line L20 through the line L18.

Hereinafter, the processes executed in the DCT processor 130 will be described in detail according to the DCT type of the processing macro block.

When DCT type information is applied, that is, the processing macro block is determined to be an object block, the DCT processing unit 130 responds to the DCT type information because the object block includes four non-transparent blocks as described in the related art. The texture information corresponding to the rearranged processing macroblock is converted into four sets of DCT coefficients, and the four sets of DCT coefficients are sequentially sent to the AC component detector 140 through the texture encoder 180 and the line L20 through the line L18. Each is delivered.

In the above, when the forward coding scheme is selected to encode the processing macroblock, the texture information of the forward block is converted into DCT coefficients in units of DCT blocks by the DCT processing unit 130, and when the forward coding scheme is selected, based on the processing macroblock. The texture information of the rearranged field block is converted into DCT coefficients in units of DCT blocks.

On the other hand, when the forward encoding designation signal is applied, the DCT processing unit 130 converts the texture information for the non-transparent block in the processing macro block determined based on the reproduction shape information into DCT coefficients.

The DCT process for the boundary block will be described with reference to FIGS. 1A to 1C. For convenience of explanation, each boundary block illustrated in FIGS. 1A to 1C is generated from reproduction shape information provided from the shape encoder 110. According to an embodiment of the present invention, since the boundary block is processed only by the progressive encoding technique, and the DCT processing unit 130 performs the DCT process on the non-transparent block, the progressive blocks P1 to P3 shown in FIGS. 1A to 1C. The texture information corresponding to is converted by the DCT processing unit 130.

In FIG. 1A, since the progressive block P1 includes only non-transparent blocks, the texture information for the processing macro block is converted into four sets of DCT coefficients in the DCT processing unit 130. On the other hand, since P2 and P3 each include two transparent blocks and two non-transparent blocks, texture information corresponding to the two non-transparent blocks is converted into two sets of DCT coefficients. As described above, it can be seen that the DCT block to be converted is selected according to the reproduction shape information.

On the other hand, if the video encoding apparatus encodes the video signal without loss, the shape information to be encoded instead of the reproduction shape information may be used in the DCT type determination process and the DCT processing process.

The DCT coefficient set generated by the DCT processor 130 is supplied to the texture encoder 180 through a line L18 and supplied to the AC component detector 140 through a line L20.

The texture encoder 180 compresses and encodes the DCT coefficient set on the line L18 using, for example, quantization, variable length encoding, or the like, thereby generating encoded texture information and providing the encoded texture information to the multiplexer 190. do.

The AC component detector 140 checks whether at least one AC component exists in the DCT coefficient set provided from the DCT processor 130 through the line L20 and provides the result to the CBPY generator 150.

As a result of the detection process in the AC component detection unit 140, if it is determined that the DCT coefficient set includes at least one AC component, the CBPY generation unit 150 determines that "1" corresponds to the DCT coefficient set on the line L20. Generate CBPY, otherwise a CBPY bit of "0" is generated. That is, one CBPY bit is allocated to a set of DCT coefficients generated from the DCT processing unit 130. If the CBPY bits for all DCT coefficient sets corresponding to the processing macro block are determined as described above, the CBPY generation unit 150 supplies CBPY information composed of the CBPY bits corresponding to the processing macro block to the CBPY encoder 160. . Referring again to FIGS. 1A to 1C, the CBPY information corresponding to the progressive block P1 includes 4 CBPY bits, and the CBPY information corresponding to P2 and P3 includes 2 CBPY bits, respectively.

The CBPY encoder 160 detects a VLC code corresponding to the CBPY information provided from the CBPY generator 150 from among the VLC tables in the VLC table unit 170, and multiplexes the detected VLC code as encoded CBPY information. To supply. Here, the VLC table unit 170 includes various VLC tables corresponding to the number of bits of CBPY information, I frames, and frame types such as P frames.

When DCT type information, coded texture information, coded CBPY information, and coded shape information are applied to the multiplexer 190 through the encoding process, the multiplexer 190 processes the data sequence illustrated in FIG. 2. Supply the data stream for the macro block to the transmission path. On the other hand, when the sequence as shown in FIG. 2 is transmitted, the decoding side can apply the same rules as the video encoding apparatus to predict the number of bits of the CBPY information using the decoded shape information.

As described above, the present invention determines the DCT type for the texture information of the video signal by using the shape information as well as the texture information of the video signal, thereby predicting the DCT type through the shape information at the decoding side, and thereby CBPY information. Can be played back without error.

Claims (2)

An image encoding an image signal including texture information and shape information for each processing macroblock that may include M * M (M is positive) pixels and may be divided into P (P is positive) DCT blocks of the same size. In the encoding device, Shape coding means for generating coded shape information by encoding shape information for the processing macroblock, and decoding the coded shape information to generate reproduction shape information; Determining the block type of the processing macroblock and providing a block designation signal indicative of the block type by using reproduction shape information from the shape coding means, wherein the block type includes only an object block including only object pixels and a background pixel; Shape checking means including a background block including and a boundary block including at least one object pixel and a background pixel; If the processing macroblock is determined as the boundary block based on the block designation signal from the shape checking means and the texture information in the video signal, the DCT type of the processing macroblock is determined by a forward coding scheme and the determined DCT type information is provided. And a DCT type determining means for determining a DCT type of the processing macro block based on texture information in the video signal and providing the determined DCT type information when the processing macro block is determined as a boundary block or an object block; And texture encoding means for encoding the texture information by a progressive encoding technique or a progressive encoding technique in response to the DCT type information from the DCT type determination means. The method of claim 1, The texture encoding means; Alignment means for rearranging the texture information and reproduction shape information in response to the DCT type information; Selecting means for detecting non-transparent blocks having a DCT block size and including at least one object pixel with respect to the processing macro block using the rearranged shape information; DCT means for converting the reordered texture information corresponding to each non-transparent block into a DCT coefficient set; Means for encoding a set of DCT coefficients corresponding to all non-transparent blocks to generate coded texture information.