KR19980074269A - Starting point coding method of contour region to be used for encoding in vertex-based shape information coding - Google Patents

Abstract

1. TECHNICAL FIELD OF THE INVENTION

Vertex-based Shape Information Coding Method

2. The technical problem to be solved by the invention

In spite of the encoding of the contour where the motion compensation region exists in the existing encoding initial point encoding, the point where the reconstructed contour and the contour of the current image coincide with the x-axis direction from the leftmost and topmost points of the VOP image. Using the absolute coordinates in the y-axis direction, the horizontal coordinates are represented by the number of bits of (log2 (number of pixels in the VOP image-1)) + 1, and the vertical coordinates are (log2 (the vertical direction of the VOP image). Since the number of pixels is equal to the number of pixels-1)) + 1, the number of bits to be encoded increases as the number of bits increases by a multiple of the number of areas, thereby reducing the coding efficiency.

3. Summary of Solution to Invention

When the same contour is stored in the receiving end and the transmitting end, if the pixel to encode the position information exists on the contour (or vertex), it is determined which pixel position of the contour (or vertex) is located, and all the contour pixels. (Or vertices) to encode the starting point bits by using the number of pixels (or the number of vertices) to represent the position of the vertices.

4. Important uses of the invention

It is applied to an inter-vertex-based shape coding scheme for efficient compression coding of contours of binary masks.

Description

Starting point coding method of contour region to be used for encoding in vertex-based shape information coding

In general, the MPEG Group, which developed and decided on international standards (MPEG-1 and MPEG-2) on video and audio coding technology and system construction, is now the next generation of video and audio, which is expected to be adopted as an international standard in November 1998. The International Standard for the Coding Technique and System Construction (MPEG-4) is being researched and developed. The development of MPEG-4 began with the need to support the next generation of video and audio applications that could not be supported by existing known standards. MPEG-4 provides new methods for communicating and accessing and manipulating video and audio data, such as object-oriented interactive functionality and access over networks of differing characteristics. In addition, it provides a characteristic that operates usefully in a communication environment where errors are easily generated and a low transmission rate communication environment. Moreover, the integration of computer graphics technology provides the ability to encode and manipulate natural and artificial images and sounds together. In summary, MPEG-4 must support many of the features required and expected in many applications. Therefore, it is possible to have an extensible and open structure to support all functions required in all low cost, high performance possible applications that are newly developed or developed by the rapid expansion and improvement of multimedia information. Among them, there is an improved compression efficiency required for transmission and storage functions and cost reduction. The present invention not only supports such a function but also can be used for an application based on various other functions.

In order to support existing applications or applications expected in the future, an image encoding technology that allows users to communicate only with desired objects in an image, access to find and read, and edit to cut and paste can be required. MPEG-4, a new video and audio encoding technology that is currently undergoing global standardization for completion in November 1998, is designed to meet this need.

The MPEG-4 standardization work under ISO / IEC is for the encoding of objects of arbitrary shape, which is the most distinguishing feature of the previous MPEG-1 and 2. For this reason, the necessity of encoding shape information has attracted attention in recent years. A representative shape information coding method considered as a standardization method is a vertex-based coding method.

The encoder used in the verification model of the conventional MPEG-4 to which the vertex-based coding scheme is applied is shown in FIG. 1.

As shown here, the switch 1 for selecting the object shape information, the shape information encoder 2 for encoding the shape information selected by the switch 1, and the shape information encoder 2 are encoded. A selection switch 3 for selecting one of the shape information and the arbitrary shape information, and a motion estimating unit for estimating the motion according to the shape information selected by the switch 1 according to the previous shape information or the shape information of the arbitrary shape. (4), the motion compensator 5 for compensating the motion of the shape information according to the motion amount estimated by the motion estimator 4, and the motion in the image obtained from the VOP forming unit (not shown). A subtractor 6 subtracting the image compensated by the compensator 5, an image signal encoder 7 encoding the image signal obtained by the subtractor 6, and a video signal compensated by the motion floating unit 5; To the video signal encoder 7 An adder 8 for adding the obtained video signal, a VOP reproducing section 9 for reproducing the VOP with the video signal obtained from the adder 8, a selection switch 3, a motion estimating section 4, and a video signal encoding section ( The multiplexer 10 multiplexes the shape information, motion information, and signal information obtained in step 7), and a buffer unit 11 for buffering and outputting the signal multiplexed by the multiplexer 10.

2 is a detailed configuration diagram of the shape information encoding unit 2, (A) is a configuration diagram at the time of intra-image encoding, and (B) is a configuration diagram at the time of inter-image encoding.

Here, the shape information encoding unit 2 during intra-image encoding, as shown in (A), displays the shape information through the switch 1 and displays the respective areas for the background, the object, and the empty areas inside the object. And an outline extracting section 2b for extracting outlines of the outer area for each region displayed by the area display section 2a, and a vertex extracting section 2c for extracting vertices to be encoded from the outlines extracted by the outline extracting section 2b. And an initial vertex encoder 2d for encoding the initial vertices among the vertices extracted from the vertex extracting unit 2c, and a vertex encoder 2e for encoding the vertices output from the initial vertex encoder 2d. The shape information encoding unit 2 at the time of the inter-image encoding, as shown in (B), displays the shape information through the switch 1 and displays the respective areas for the background, the object, and the empty areas inside the object. And the outline of the outer area for each area displayed on the area display section 2a. A contour extractor 2b for extracting the data, a contour predictor 2f for predicting the contour information of the current image extracted by the contour extractor 2b using previous shape information, and transmitting only an error region thereof, and a contour prediction. A vertex extracting unit 2c for extracting vertices to be encoded from the contour information obtained from the section 2f, an initial vertex encoding unit 2d for encoding initial vertices among the vertices extracted from the vertex extracting unit 2c, and an initial vertex It consists of the vertex encoding part 2e which codes the vertex output from the coding part 2d.

The vertex-based technique, which is one of the shape information encoding techniques under consideration in the current MPEG-4, is constructed as follows. Unlike conventional video encoding international standards such as MPEG-1, MPEG-2, H.261, and H.263, MPEG-4 encodes only meaningful objects in an image and freely encodes objects in the image by encoders and decoders. Techniques for accessing and editing code are standardized.

Accordingly, when an image including a meaningful object enters the MPEG-4 encoder together with the shape information as shown in FIG. 3A, the shape information encoder uses the area display unit 2a of FIG. 2 as shown in FIG. 3A. As shown in (b) to (e) of FIG. 3, a meaningful object in the image is displayed for each region by the background, the object, and the empty area inside the object. For each area displayed by the area display unit 2a, the outline of the outline is extracted by the outline extracting unit 2b, and the outline information of each area in the image obtained is obtained from the vertex extracting unit 2c and the initial vertex encoding in the image encoding. 2d, the vertex encoder 2e encodes the motion vector to be encoded if the prediction is successful after predicting the previous shape information by the contour predicting unit 2f during inter-image encoding. After skipping the contour encoding process, if the prediction fails, the contour information of the current image is output to the vertex extracting unit 2c, and the vertex extracting unit 2c, the initial vertex encoding unit 2d, and the vertex encoding unit 2e are used. Encode

In FIG. 3, (a) is input shape information, (b) shows a background area, (c) shows an area of object 1 (contour line type = object), and (d) shows an area of object 2 And (contour type = object), (e) is the area mark of object 3 (contour type = empty area in the object).

In shape information encoding, in-image encoding that encodes only shape information of the current image when encoding shape information, and predicts current contour information by using contour information of a previous image, and does not encode an area in the contour that exists within the marginal error area. Instead, there is a method for inter-image encoding of a contour that encodes an area within the contour that exists outside the error region.

After the motion estimation is performed in the encoding of the shape information between the vertices, the contour information of the object 1 of the previous image is used to predict the contour of the object A of the current image as shown in FIG. 4, and the object 2 of the previous image is Used to predict object B, and the contour of object C of the current image is determined to be the intra-image coding mode because the contour that can be used for prediction in the previous image is determined, the object A contour of the current image based on the contour of the current image In this case, the object B contour is encoded using the object contour of the previous image, and then intra-image coding is applied to the remaining regions except for the contour region that can be encoded between images, and the object C contour is not a predictable contour. Apply. When encoding the object A contour and the object B contour, inter-image coding should be performed on the areas where the predicted contour after the moving compensation and the contour of the original image exceed the allowable error. The location information of the encoding initial point to be used as a reference point for the encoding should be encoded.

At this time, the existing method of encoding the encoding initial point is the absolute coordinate of the x-axis direction and the y-axis direction from the VOP image, the leftmost side, and the topmost point where the previously reconstructed contour predicted as the encoding initial point and the contour of the current image match. By using the values, the horizontal coordinates are expressed as the number of bits of (log2 (pixels in the horizontal direction of the VOP image-1)) + 1, and the vertical coordinates are (log2 (number of pixels in the vertical direction of the VOP image-1)). The number of bits is +1.

However, although the encoding method of the initial point of encoding is the encoding of the contour where the region capable of motion compensation exists, the x-axis from the leftmost and uppermost endpoints of the VOP image corresponds to the point where the reconstructed contour matches the contour of the current image. By using the absolute coordinate values in the direction and y-axis direction, the horizontal coordinates are represented by the number of bits of (log2 (number of pixels in the horizontal direction of the VOP image-1)) + 1, and the vertical coordinates are represented by (log2 (the Since the number of pixels to be encoded in the vertical direction is represented by the number of pixels-1)) + 1, the more the number of regions to be encoded, the more the number of bits increases by a multiple of the number of regions, thereby reducing the coding efficiency.

Accordingly, the present invention has been proposed to solve various problems that occur when encoding the contour region whose prediction fails in the vertex-based shape information encoding as described above. In vertex-based shape information encoding to determine the point where the contour of the original image coincides with the surrounding vertices as the starting point of the region to be encoded, and allocate the bits according to the number of pixels on the contour or the number of vertices on the vertices. The present invention provides a method of encoding a starting point of an outline region to be used for encoding.

The method for achieving the object of the present invention, when the same contour is stored in the receiving end and the transmitting end, if the pixel to be encoded the position information is located on the contour, it determines the position of the pixel position of the contour, and In order to represent the positions of all the contour pixels, the start point bits are assigned and encoded using the number of pixels.

Another method for achieving the object of the present invention, when storing the same contour at the receiving end and the transmitting end, if the pixel to be encoded the position information is located on the vertex, it determines the position of the pixel position of the vertex, and In order to represent the positions of all the vertices, the vertex number is allocated and encoded using the starting point bits.

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

1 is a block diagram of a conventional MPEG-4 encoder to which the present invention is applied;

2 is a detailed configuration diagram of a vertex-based shape information encoder;

(A) is a detailed block diagram of the shape information encoder at the time of intra-image encoding;

(B) is a detailed configuration diagram of the shape information encoding unit at the time of encoding between images;

3 is a view illustrating an area display of the area display unit of FIG. 2;

Figure 4 is an example of the image configuration for explaining the present invention.

FIG. 5 is an example diagram of motion compensated contours of shape information of a previous image and contours of shape information of an original image when there is a contour region to which inter-image encoding is applied; FIG.

* Description of the symbols for the main parts of the drawings *

2: shape information encoding unit 2a: area display unit

2b: contour extraction unit 2c: vertex extraction unit

2d: initial vertex encoder 2e: vertex encoder

2f: contour prediction unit

The present invention is a technique applied to the contour predictor, the vertex extractor, the initial vertex encoder, and the vertex encoder of FIG.

Accordingly, the present invention refers to the above-described technical configuration, and when the same contour is stored in the receiving end and the transmitting end, if the pixel to be encoded is located on the contour, it is determined which pixel position of the contour is located. In this case, the start point bits are allocated by using the number of pixels so as to represent the positions of all the contour pixels.

In the present invention, when the same contour is stored at the receiving end and the transmitting end, when the pixel to encode the position information is located on the vertex, it is determined which pixel position of the vertex is located, so that the positions of all vertices can be expressed. By using the number of vertices is characterized in that the start point bits are assigned and encoded.

Hereinafter will be described the operation of the preferred embodiment of the present invention.

When the inter-image encoding technique of the contour is applied, regions smaller than the allowable error and the allowable error between the contour of the motion compensated contour of the previous reconstructed image by the estimated motion vector and the contour of the current image as shown in FIG. It is divided into areas larger than the error area. For example, between A and B is an area larger than the allowed error, between B and C is smaller than the allowed error, between C and D is greater than the allowed error, and between D and A is allowed. The area becomes smaller than the error. In the inter-image encoding of the contour, the areas A to D and B to C are not encoded, and the areas A to B and C to D are encoded in the image or error information is encoded. In order to encode areas A to B and areas C to D, an initial point when encoding the area must be determined, and position information of the vertex must be encoded. The existing technology currently used is determined from two points A and B where the contour of the motion-compensated previous reconstructed image and the current contour coincide, and the position information of the point uses absolute coordinate values from the leftmost and top pixels of the VOP image. To encode it.

Since the VOP width and height are known, (log2 (VOPwidth-1) +1) + (log2 (VOP height-1) +1) bits are required to encode the position information.

However, in the present invention, since both the encoder and the decoder have the contour information of the previous reconstructed image, the point (A, B, C, D) where the two contours coincide is used as an initial vertex to start encoding or vertex around it. (A, B, C, D) can be defined as the initial vertex to start coding. When determining the point on the contour as the encoding initial vertex, the positional information may be determined according to the number of pixels on the contour, and when determining the vertex around the point on the contour as the encoding initial point, the positional information may be determined according to the number of vertices. . In other words, encoding is possible only by transmitting information indicating the number of pixels from a predetermined start point.

The present invention is a technique using such a point, wherein the number of bits required for encoding requires log2 (number of pixels on the contour-1) + 1 bit when using the point on the contour of the reconstructed previous image, and vertices around the point. When using, log2 (number of vertices-1) +1 bit is needed. Therefore, when using the existing technology, assuming that the VOP size is 176 × 145, the number of pixels on the contour of the reconstructed previous image is 350 and the number of vertices is 40, 16 bits are required when the existing technology is applied. The technique requires 9 bits to set the point on the contour as the initial vertex, and 6 bits to set the vertex around the point as the initial vertex. As an example, the bit reduction effect of 5 bits or 7 bits can be seen only at the time of initial vertex encoding. However, since there may be more than one region to be encoded as shown in FIG. 5, (5 or 7 bits) * contour The bit reduction effect can be seen as much as the number of regions to be encoded in the frame.

Therefore, as the number of regions to be encoded in the contour of the previous image and the current image increases, the encoding efficiency may be further improved.

As described above, according to the present invention, when the region to be encoded is encoded during inter-image encoding in vertex-based shape information encoding, a bit representing the start point of the region to be encoded is log2 (number of pixels on the outline-1) +1 bit or Since only log2 (number of vertices-1) + 1 bit is required, the starting point positioning bits can be reduced by the number of regions to be encoded in the contour, thereby increasing the coding efficiency.

In addition, the more regions to be encoded, the more the bit is reduced by a multiple of the number of regions, thereby improving the coding efficiency.

The present invention determines the point where the predicted contour and the contour of the original image coincide with each other and the surrounding vertices as the starting point of the region to be encoded, and allocates bits according to the number of pixels on the contour or the number of vertices on the vertices. The purpose of the present invention is to provide a method of encoding a starting point of a contour region to be used for encoding in vertex-based shape information encoding to improve efficiency of contour coding.

Claims (2)

In the method of determining the starting point of the region to be encoded when encoding the contour information between the vertex-based shape information encoding method and encoding the position information, When the same contour is stored at the receiving end and the transmitting end, if the pixel to encode the position information exists on the contour, it is determined by which pixel position of the contour, and the number of pixels to express the position of all contour pixels. A start point encoding method of a contour region to be used for encoding in vertex-based shape information encoding, characterized by assigning and encoding a start point bit. In the method of determining the starting point of the region to be encoded when encoding the contour information between the vertex-based shape information encoding method and encoding the position information, When the same contour is stored in the receiver and the transmitter, if the pixel to encode the location information is located on the vertex, it is determined by which pixel position of the vertex it is located and the number of vertices is used to express the position of all vertices. And a starting point bit is allocated and encoded, wherein the starting point encoding method of the contour region to be used for encoding in the vertex-based shape information encoding.