KR100289698B1 - Method and apparatus for coding object information of image object plane - Google Patents

Abstract

PURPOSE: A method for coding object information of a VOP(Video Object plane) is provided to add each object symbol to a background pixel included in a boundary block to generate processing blocks, and to code more than one boundary blocks including object blocks and object symbols as well as coding processing blocks including unprocessed boundary blocks, to generate coded video signals, so as to improve coding efficiency of video signals having large object information. CONSTITUTION: A block sensor(332) separates video images into many blocks, and sorts out each block to one of background block, boundary block, and object block. The background block includes a background pixel only. The boundary block includes an object and a background pixel. The object block includes an object pixel only. A randomize object data generator(336) generates binary object symbol sets presenting object information. A processing block generator(338) couples object symbols with the object and the boundary blocks to generate processing blocks. A conversion unit(334) performs a conversion-coding for the processing blocks to generate coded object information.

Description

Object information encoding method and apparatus of an image object plane {METHOD AND APPARATUS FOR CODING OBJECT INFORMATION OF IMAGE OBJECT PLANE}

The present invention relates to a method and apparatus for encoding object information of an image object plane, and more particularly, to a method and apparatus for encoding object information of an image object plane to improve encoding efficiency.

In digital television systems such as video telephony, teleconferencing, and high-definition television systems, a video line signal within a video frame signal consists of a series of digital data called pixel values, so a large amount of digital data is needed to define each video frame signal. Is needed. However, due to the limited frequency bandwidth available for common transmission channels, especially for low bit rate video signal encoders such as video telephony and teleconferencing systems, in order to transmit large amounts of digital data there are various data compression techniques. Through compression, it is inevitable to compress or reduce the amount of data.

One of the techniques for encoding video signals in such a low rate coding system is the so-called object-oriented analysis-synthesis coding technique (MPEG-4 Video Verification Model Version 7.0, International Organization for Standardization, Coding of Moving Pictures and Associated Audio Information). , ISO / IEC JTC1 / SC29 / WG11 MPEG07 / N1642, Bristol, April 1997). According to the object-oriented analysis-synthesis coding technique, an input video image is divided into objects, that is, a video object plane (VOP), which corresponds to an entity in a bit stream that a user can access and process. ; Three sets of variables for defining the motion, shape, and texture information of each object are processed through different coding channels.

A VOP, called an object, is a bounding rectangle that is the minimum multiple of 16 pixels in width and height surrounding each object, and the encoder processes the input image image in VOP units, that is, in units of objects. The VOP includes texture information indicating contrast and color information of the object and shape information indicating the shape and position of the object.

In texture information, a pixel is represented by one of values between 1 and 255, for example; In the shape information, for example, binary 0 is used to represent a pixel outside the object of the VOP, that is, a background pixel, and another binary 255 is used to represent a pixel inside the object, that is, an object pixel.

Referring to FIG. 1, a general encoding apparatus 100 for encoding texture information of a VOP is illustrated. The converter 105 converts the object pixel value of the shape information of the VOP to 1 and provides the converted shape information to the multiplier 110 and the block selector 125. The converted shape information has a value of 1 And the value of the background pixel is zero. In the multiplier 110, the texture information of the VOP is multiplied by the converted shape information to generate the converted texture information and are provided to the padding unit 120. Object pixels in the converted texture information retain their original values, and pixels outside the object have a value of zero.

The padding unit 120 performs a padding process on the background pixels in the converted texture information by using a general repetitive padding technique. In the repetitive padding technique, the background pixels are applied to the boundary pixel value of the object. Padded with new values obtained on the basis. Referring to FIG. 2, a VOP 10 represented by padded texture information is shown. The VOP 10 includes an object 15 displayed as a hatched portion and a background 20 displayed as a non-hatched portion. Include. The block selector 125 detects the object 15 and the background 20 in the VOP 10 according to the converted shape information and the padded texture information; The VOP 10 is divided into a number of discrete cosine transform (DCT) blocks, for example 8 × 8 pixels, such as 1-12. Then, the block selector 125 selects DCT blocks overlapping the object 15 and provides them to the DCT unit 130 as processing blocks.

In the example shown in FIG. 2, DCT block 1 does not overlap 15 with the object; Thus, only two to twelve of the DCT blocks are selected as processing blocks. In the DCT unit 130, each processing block is converted into a set of DCT coefficient sets and the DCT coefficient set is provided to the quantization unit 140. In quantization unit 140, a set of DCT coefficients is quantized and sent to a transmitter (not shown) for transmission.

On the other hand, object information, for example, the index, title, author, user-editability, etc. of the object is encoded separately from the movement, shape, and texture information of the object and transmitted as a header of the encoded sequence of the object. . Therefore, when the number of objects to be transmitted is large, the amount of object information is increased, thereby reducing the coding efficiency.

Accordingly, an object of the present invention is to provide a method and apparatus for effectively encoding object information of a video object plane (VOP).

In order to achieve this object, according to an aspect of the present invention, in a method of encoding a video signal including a video object plane (VOP) and its object information, the VOP may include an object therein. When: (a) Divide the VOP into a number of blocks and detect the boundary blocks and object blocks, each boundary block contains both background pixels and object pixels, and the object blocks contain only object pixels, and background and object The pixels are pixels belonging to the outside and inside of the object, respectively; (b) converting object information represented by a binary number into an object symbol set; (c) add each object symbol to a background pixel contained in the boundary block to generate a process block, the process block remaining without including one or more processed boundary blocks and object symbols comprising the object block and the object symbol; Including unprocessed boundary blocks; And (d) encoding the processing block to generate encoded object information.

In order to achieve this object, according to another aspect of the present invention, in the apparatus for encoding a video signal including a video image including the object and the object information, the video image is composed of object pixels and background pixels, When the object pixel exists inside the object and the background pixel exists outside the object: The image image is divided into a plurality of blocks, and each block is classified into one of a background block, a boundary block, and an object block. Video image separation means including only pixels, the boundary pixels including objects and background pixels, and the object blocks including only object pixels; Object symbol set generating means for generating a binary object symbol set representing object information; Processing block generating means for combining the object symbol with the object block and the boundary blocks to generate processing blocks; And means for transform encoding the processing blocks to generate encoded object information.

1 is a block diagram of a general encoding apparatus for encoding information of a video object plane (VOP),

2 illustrates a VOP divided into a plurality of DCT blocks;

3 is a block diagram of an apparatus for encoding object information according to a preferred embodiment of the present invention;

FIG. 4 is a detailed view of the object information insertion unit shown in FIG. 3.

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

1: Background block 2-6, 8-12: Boundary block

7: object block 10: VOP

15: Object 20: Background

305: converter 310: multiplier

320: padding unit 330: object information insertion unit

332: block detection unit 334: conversion unit

336: randomized object data generation unit

338: processing block generation unit 340: conversion unit

350: quantization unit 360: modulus unit

370: VLC part

3, there is shown a block diagram of an apparatus 300 for encoding object information in accordance with a preferred embodiment of the present invention. The functions and characteristics of the transducer 305, multiplier 310 and padding 320 are the same as the corresponding elements 105, 110 and 120 shown in FIG. 1 and are not described again for convenience. The padded texture information from the padding unit 320 and the converted shape information from the converter 305 are applied to the object information inserting unit 330 that receives the object information. A detailed process of encoding the object information by the object information inserting unit 330 will be described with reference to FIGS. 2 and 4, and FIG. 4 shows a detailed block diagram of the object information inserting unit 330.

Similar to the padding unit 120 illustrated in FIG. 1, the block detector 332 detects a DCT block overlapping the object 15. Among the DCT blocks in the VOP 10, DCT block 1 is defined as a background DCT block because it does not overlap with the object 15 and is composed of only background pixels, and DCT block 7 is determined as an object block composed only of object pixels. Blocks 2 through 6 and 8 through 12 are boundary DCT blocks including both object pixels and background pixels. After detecting the object and background DCT blocks, the block detector 332 attaches an identification flag to each background pixel, which indicates that the pixel to which the flag is attached belongs to the background 20. The object and background DCT blocks are provided to the processing block generation unit 338 as processing DCT blocks.

On the other hand, the object information is input to the conversion unit 334. Each letter, number, symbol or the like representing the object information in the conversion unit 334 is converted into a binary symbol represented by an 8-bit binary number, for example. The binary symbol set generated by the conversion unit 334 is applied to the randomized object data generation unit 336. The set of binary symbols forms a constant pattern, which is very different from the pattern of the object pixels, resulting in lower coding efficiency of subsequent transform fixation, eg, DCT. For this reason, the randomized object data generator 336 generates a randomized object symbol set by multiplying each binary symbol by a random number.

The random number is selected from among a set of random numbers preset in a predetermined manner. The random number should be large enough so that the object information is not damaged in subsequent quantization. The randomize object symbol set from the randomize object data generator 336 is applied to the processing block generator 338. The processing block generation unit 338 detects the background pixels in the processing DCT block based on the identified identification flag.

In general, blocks are processed in raster scanning order from left to right and top to bottom. Therefore, the processing DCT blocks shown in Fig. 2 are processed in the order of DCT blocks 2 to 12. The processing block generation unit 338 finds the first processing DCT block, that is, block 2, and adds the randomized object symbols and the padded background pixel values selected in the raster scanning order in the DCT block 2 one by one. If a randomized object symbol has not yet been added, these object symbols are added to the padded background pixel of the boundary DCT block followed in the manner described above until all the randomized object symbols have been processed. When a randomized object symbol is processed, one or more processed boundary DCT blocks contain the randomized object symbol information, and the remaining unprocessed boundary DCT blocks and all object DCT blocks remain intact.

Referring back to FIG. 3, the transform unit 340 transforms each processing block into a transform coefficient set by using a general discrete cosine transform (DCT) technique, and provides it to the quantization unit 350. Here, the transform coefficient set is converted into a quantized transform coefficient set and provided to the modulus unit 360.

In the modulus unit 360, a modulus process is performed on pixels exceeding a predetermined maximum value based on a general modulus technique. The range of quantization transform coefficient values is limited within a preset maximum value by subtracting an integer multiple of the preset maximum value, ie modulus, where the modulus is, for example, 255. For example, in the subsequent statistical coding process based on a variable length coding (VLC) technique, the maximum value is 255, and if the value of the quantization transform coefficient corresponding to the background pixel including the randomized object symbol is 520, the quantization transform coefficient value Decreases to 10 (= 520-255 x 2), and modulus information is appended, which indicates the number of modulus processes.

On the other hand, if the value of the quantization transform coefficient corresponding to the background pixel including the randomized object symbol is 270, the quantization transform coefficient value is 15 (= 270-255 x 1), and the modulus information is 1. In the VLC unit 370, the modulated data is VLC encoded and sent for transmission to a transmitter (not shown).

In the decoder on the receiver side, the transmitted VLC coded data is reproduced through a series of variable length decoding, inverse modulus, inverse quantization, and inverse transformation. Thus, texture information of the object is obtained from the processed processing blocks and the shape information of the object transmitted from the transmitter. Then, an inverse padding process is performed based on the boundary pixels of the texture information of the reproduced object and the padded pixel values of the background pixels are removed. At this time, the remaining values of the boundary pixel are divided by preset random number sets previously stored in the decoder, so that a binary symbol corresponding to the object information is generated. These binary symbols are translated into object information.

While specific embodiments of the invention have been described above, those skilled in the art can make various changes without departing from the scope of the claims described herein.

As described above, when the VOP includes an object, the VOP is divided into a plurality of blocks, and when the background pixel is a pixel located outside the object and the object pixel is a pixel located inside the object, only the background pixel is used. Finds the background block and the object block including only the object pixel and the boundary block including both the background pixel and the object pixel, converts the object information into the object symbol set, and converts each object symbol to the background included in the boundary block. The amount of object information is generated by generating processing blocks in addition to the pixels, and then encoding the processing blocks including the object block and one or more processed boundary blocks containing the object symbols and the remaining raw boundary blocks to produce an encoded image signal. The coding efficiency of many video signals can be improved.

Claims (10)

In a method for encoding a video signal comprising a video object plane (VOP) and its object information, the VOP includes an object therein: (a) dividing the VOP into a plurality of blocks and detecting boundary blocks and object blocks, each boundary block including both a background pixel and an object pixel, the object block including only an object pixel, and a background and object pixel Are pixels belonging to the outside and inside of the object, respectively; (b) converting object information represented by a binary number into an object symbol set; (c) adding each object symbol to a background pixel contained in the boundary block to generate a processing block, wherein the processing block remains without one or more processed boundary blocks and object symbols including the object block and the object symbol; Including unprocessed boundary blocks; And (d) encoding the processing block to generate an encoded object information signal. The method of claim 1, wherein the converting step (b) is: (b1) converting object information into a set of binary symbols, wherein each binary symbol is represented by a P bit which is a positive integer; And (b2) generating an object symbol set by multiplying each binary symbol by a random number selected from a set of preset random numbers. The method of claim 1, wherein the encoding step (d) is: (d1) transforming each processing block to provide a transform coefficient set; (d2) quantizing the transform coefficient set to generate a quantized coefficient set; And (d3) encoding the quantization coefficient set based on a statistical encoding technique to generate an encoded image signal. The method of claim 1, wherein the adding step (c) is: (c1) detecting one or more boundary blocks and combining object symbols therein; And (c2) generating the processed boundary blocks by adding each object symbol to the background pixel included in the detected boundary blocks. An apparatus for encoding a video signal including a video image including an object and object information, wherein the video image includes object pixels and background pixels, the object pixel inside the object, and the background pixel outside the object. When each exists: The image is divided into a plurality of blocks, and each block is classified into one of a background block, a boundary block, and an object block, wherein the background block includes only background pixels, the boundary pixels include objects and background pixels, and the object blocks are objects. Video image separation means including only pixels; Object symbol set generating means for generating a binary object symbol set representing object information; Processing block generating means for combining the object symbol with the object block and the boundary blocks to generate processing blocks; And And means for transform-coding the processing blocks to generate encoded object information. 6. The apparatus of claim 5, wherein the means for generating an object symbol set are: Means for converting the object information into a set of binary symbols, wherein each binary symbol is represented by a predetermined number of binary digits; And And means for obtaining an object symbol set by multiplying each binary symbol with a random number selected from a preset random number set in a predetermined method. 6. The object information of claim 5, wherein the processing blocks comprise one or more processed boundary blocks including object blocks and object symbols and boundary blocks remaining without object block and object information. Encoding device. 6. The processing block according to claim 5, wherein the processing block generating means is: Means for finding one or more boundary blocks to combine the object symbols; And And means for generating processed boundary blocks by adding each object symbol to a background pixel included in the found boundary blocks. The object of claim 5, wherein the image image separating means attaches an identification flag to background pixels included in the boundary blocks, and an identification flag attached to one pixel indicates that the pixel is a boundary pixel. Information encoding apparatus. 10. The apparatus of claim 9, wherein the background pixel to which one or more boundary blocks and each object symbol are added is detected by an identification flag.

Images