JPH10336673A - Edge detection method and its device in video signal coding system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect an edge of a boundary tetragonal that surrounds a video object plane in crossing with contour lines of an object by combining contour line data and additional contour line information so as to generate a closed contour line of a prescribed border rectangle. SOLUTION: A cross point detection section 310 detects whether or not a detected contour line is in crossing with an edge of a border rectangle surrounding a current video object plane(VOP) based on contour lie information on a line L10 and the current VOP on a line L 50. A switch 332 is used to give the current VOP on the line L 50 to a start point/end point decision section 312, a side selection section 314, and a position information generating section 316 or an optimum contour line decision section 350 selectively based on a detection signal from the cross point detection section 310. A buffer 318 extracts all position information in response to a read control signal from a control section 330 and gives the information to a contour line generating section 320. The contour line generating section 320 combines position information in the buffer 318 with received contour line information to form a new closed contour line in the boundary rectangle.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a contour coding system for coding a contour of an object represented by a video signal, and more particularly to an image which intersects a contour of the object within a predetermined boundary rectangle. The present invention relates to a method and an apparatus for detecting an edge of a corresponding bounding rectangle surrounding an object plane (VOP).

[0002]

2. Description of the Related Art Generally, in a digital video system such as a videophone or an electronic conference, a video frame signal is composed of a series of digital data called "pixels". Of digital data is required. However, the available frequency bandwidth on a regular transmission channel is limited, so transmitting large amounts of digital data over that channel is especially important for video transmission at low bit rates such as video telephony and teleconferencing. For signal coding systems, various data compression techniques must be used to compress or reduce the amount of data to be transmitted.

[0003] In a video signal encoding system of a low bit rate, one of the methods of encoding a video signal is a so-called video signal encoding method.
Object-oriented analysis / synthesis coding method (Object-orien
tedanalysis-synthesis coding technique). According to this object-oriented analysis / synthesis coding technique, an input video is divided into a plurality of objects (objects), and an encoding channel having three sets of parameters that define the motion, outline, and pixel data of each object is different. Is handled through.

As an example of this object-oriented encoding method, a so-called MPEG-4 (Moving Picture Expert) is used.
s Group-4), but this technique provides a standard for audiovisual coding that allows content-based interaction, low bit rate communication, interactive multimedia (eg, games, interactive TV, etc.). And to improve coding efficiency and / or general access capabilities in application areas such as mechanisms for area monitoring (eg, MPEG
-4 Video Verification Model Version 2.0, Intern
ational Organization for Standardization, ISO / IEC
JTC1 / SC29 / WG11 / N1260, March, 1997).

According to MPEG-4, an input video frame is divided into a plurality of video object planes (VOPs) corresponding to entities accessible or manipulated by a user in a bitstream.

A VOP is called an object, and surrounds each object and has a width and a height of 16 pixels (the size of a macroblock).
May be represented as a bounding rectangle that is the smallest multiple of. Therefore,
The encoder processes the input video image on a VOP basis, that is, on an object basis. That is, each VOP is represented by a predetermined boundary rectangle, and the phase difference between the luminance data Y and the chrominance data (U, V) of the boundary rectangle is 4: 2: 0 as shown in FIG. It should be determined exactly according to the format. Luminance data and chrominance data are represented by x and o, respectively. Specifically, as shown in FIG. 2, first, in the absolute coordinate system (frame), the coordinates of the upper left corner of the bounding rectangle 10 are the coordinates of the upper left corner of the minimum rectangle 20 surrounding the object 30, for example, (2n
+1, 2m + 1) and should be set to the nearest even coordinate, eg, (2n, 2m). Where n
And m are each a positive integer. Then, the corner at the lower right corner of the boundary rectangle 10 is expanded so that the width and height of the rectangle represented by the chrominance data are the minimum multiple of 16 pixels. Therefore, the coordinates of the upper left corner of the rectangle represented by the chrominance data are the coordinates of the luminance data divided by two.

[0007] The VOP disclosed in MPEG-4 has shape information and color information composed of luminance data and chrominance data. The shape information is represented by, for example, a binary mask and is related to the luminance data. In the binary mask, one binary, eg, “0”, is used to represent a pixel located outside the object in the VOP, as shown in FIG. 3, and the other binary, eg, “1”, Used to represent pixels located inside the object.

On the other hand, in the VOP encoder,
The motion of the current contour included in the OP is estimated. More specifically, the optimum contour most similar to the current contour is detected from the predicted contours existing in the predetermined search area.
After detecting the optimal contour, a motion vector representing the displacement between the current contour and the optimal contour and index data of the optimal contour are obtained. After that, a predicted current contour is obtained based on the motion vector and the index data,
Also, using a normal contour inter-coding technique, inter-coding is performed on the current contour and the predicted current contour together with the motion vector and index data.

However, in the VOP encoder, if a part of the current contour is located outside the boundary rectangle and intersects the edge of the screen or video frame, the optimal contour that is optimally matched with the current contour in the VOP is determined. Hard to find.
Therefore, in such a case, the edge of the screen is treated as a part of the current contour, and a new contour including that part and the contour of the object within the bounding rectangle is used to determine the optimal contour. . However, an apparatus or method for detecting an edge of a screen that intersects a current contour is not yet known.

[0010]

SUMMARY OF THE INVENTION Accordingly, a primary object of the present invention is to detect, within a predetermined bounding rectangle, the edge of the bounding rectangle surrounding the VOP that intersects the contour of the object. It is an object of the present invention to provide a method and an apparatus therefor.

[0011]

According to a preferred embodiment of the present invention, there is provided a video object plane (VOP) which intersects a contour of an object represented in a video signal.
Edge detection method for detecting an edge of a predetermined boundary rectangle surrounding the predetermined boundary rectangle, the method including receiving VOP data having outline data representing the position of the outline pixel of the VOP and information relating to the predetermined boundary rectangle. Detecting whether or not the contour of the VOP data intersects the edge of the predetermined boundary rectangle, and determining the intersection, the intersection is determined on the edge of the predetermined boundary rectangle where the contour intersects. A, and based on the contour data, the intersection and the information of the boundary rectangle, sequentially obtain position information of a plurality of points located along the side of the boundary rectangle between the intersections, Based on the information, an additional contour line corresponding to a portion consisting of a plurality of points located between the intersection points along a corresponding side of the predetermined boundary rectangle corresponding to the intersection point is obtained, and the corresponding additional contour is obtained. Line to line C to generate additional contour information
An edge detection method, comprising: generating a closed outline within the predetermined boundary rectangle by combining the outline data and the additional outline information. .

According to another embodiment of the present invention, a video object plane (VO) which intersects with the contour of the object represented in the video signal.
An edge detection device for detecting an edge of a predetermined boundary rectangle surrounding P), wherein the VOP data includes contour data indicating a position of a contour pixel of the VOP and information relating to the predetermined boundary rectangle. To detect whether or not the contour line of the VOP data intersects with the edge of the predetermined boundary rectangle, and when it intersects, detects an intersection on the edge of the predetermined boundary rectangle where the contour line intersects Based on the contour data, the intersection, and the boundary rectangle information, sequentially determine the position information of a plurality of points located along the side of the predetermined boundary rectangle between the intersections. Obtaining, based on the position information, obtaining an additional contour line corresponding to a portion consisting of a plurality of points located between the intersections along a corresponding side of the predetermined boundary rectangle corresponding to the intersection. Contour Additional contour line information generating means for generating additional contour line information for a line; closed contour line generating means for generating a closed contour line within the predetermined boundary rectangle by combining the contour line data and the additional contour line information An edge detection device is provided, comprising:

[0013]

Preferred embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 4 is a schematic block diagram of an apparatus 900 for encoding a VOP expressed in a video signal according to the present invention. The VOP is the current VOP, and the contour detection unit 10
0 and the motion estimation unit 3
00 is input in the form of a division mask. The current VOP has outline data indicating the position of the outline pixel, and the position and size information of a predetermined boundary rectangle surrounding the current VOP. Each pixel in the division mask has a label for identifying a region to which the pixel belongs. For example, a pixel located in the background has a label of “0” and the current VO
Each pixel in P is labeled with one of the non-zero values.

The contour detection unit 100 detects a contour from the current VOP, assigns an index to the detected contour, and outputs contour data including contour data and index data for the contour. After that, the contour information is transferred to the inter-encoding unit 200 through the line L10.
And the motion estimation unit 300. Motion estimation unit 300
Is the outline information on the line L10 and the frame memory 70
Based on the previously reconstructed contour video signal supplied from 0 through line L30, the contour most similar to the contour in the current VOP is detected. Here, the previously reconstructed contour image signal is input in a divided mask form,
Each pixel included in the division mask has a label indicating a region to which the pixel belongs. The motion estimating unit 300 outputs the index data of the most similar contour and the motion information indicating the displacement between the contour in the current VOP and the most similar contour to the lines L20 and L40, respectively. Motion estimation unit 3
00 will be described in detail below with reference to FIGS. 5, 6A and 6B.

FIG. 5 is a detailed block diagram of the motion estimator 300. The intersection detector 310 and the start point / end point determiner 31 are provided.
2, a side selection unit 314, a position information generation unit 316, a closed outline generation unit 320, a control unit 330, and an optimum outline determination unit 350.

The intersection detection unit 310 determines the current VO based on the outline information on the line L10 and the current VOP on the line L50.
A detection signal is generated by detecting whether or not the contour detected from P intersects the edge of the bounding rectangle surrounding the current VOP. For example, as shown in FIG. 6A, when an intersection (for example, A, B) exists between the contour 115 and a predetermined boundary rectangle, the intersection detection unit 310 generates a high-level detection signal, and If not, a low-level detection signal is generated and supplied to the switch 332 and the start / end point determination unit 312.

The switch 332 starts the current VOP on the line L50 through the line L11 in response to the high or low level detection signal received from the intersection detection unit 310.
It is supplied to the end point determining unit 312, the side selecting unit 314, and the position information generating unit 316, or is selectively supplied to the optimal contour determining unit 350 via the line L12. More specifically, when a low-level detection signal is input, the current VOP is supplied to the optimum contour determining unit 350 through a line L12, and when a high-level detection signal is input, the current VOP starts at a line L11. The information is transmitted to the end point detection unit 312, the side selection unit 314, and the position information generation unit 316.
As shown in FIG. 6B, the start / end point detection unit 312 responds to the detection signal from the intersection point detection unit 310 by using the switch 332.
, The intersections A and B on the current VOP supplied through the line L11 are respectively assigned as a start point and an end point. According to the embodiment of the present invention, the intersection A that first appears counterclockwise from the inside to the outside of the boundary rectangle is determined as a start point, and the intersection B that appears last is determined as an end point. After that, the start point / end point detection unit 312 compares the position information of the start point and the end point with the side selection unit 314 and the position information generation unit 316.
Respectively.

Thereafter, as shown in FIG. 6B, the side selector 314 receives the position information indicating the starting point A from the controller 3.
Under the control of 30, one of the four sides 111 to 114 of the boundary rectangle of the current VOP is selected (for example, the upper side 111), and information representing the upper side 111 is transmitted to the position information generation unit. 316. As shown in FIG. 6B, the position information generation unit 316 determines the start point A based on the current VOP supplied from the switch 332 through the line L11, the outline information on the line L10, and the side information from the side selection unit 314. It starts to generate position information of all points between and the coordinates (Xmin + 1, Ymin) on the upper side 111 of the bounding rectangle. As can be seen from the above, such position information can be easily obtained based on the outline information and the position and size information of the bounding rectangle. Thereafter, the position information is supplied to the control unit 330 and is also supplied to the buffer 318 to be temporarily stored.

The control section 330 controls the operation of the side selection section 314 based on the position information from the position information generation section 316. More specifically, the control unit 330 controls the position information generation unit 3
When all the position information between the start point A and the coordinates (Xmin + 1, Ymin) is received from 16, the side selection unit 314 selects the side connected counterclockwise from the upper side 111 of the bounding rectangle, that is, the left side 112. Control signal.

The side selector 314 selects the left side 112 of the bounding rectangle in response to a control signal from the controller 330, and outputs side information corresponding to the same to the position information generator 3.
16. Similarly, the position information generation unit 316 starts outputting position information of all points between the coordinates (Xmin, Ymin) and the end point B on the left side 112 of the boundary rectangle according to the side information. The position information is stored in the control unit 33
0 and is also supplied to buffer 318 and stored temporarily. When the control unit 330 receives the position information of all points between the start point A and the end point B, it generates a read control signal for controlling the operation of the buffer 318. The buffer 318 extracts all the stored position information according to the read control signal from the control unit 330 and supplies it to the closed contour generation unit 320. For convenience of explanation, the start point A and the end point B, which are located only on the two sides 111 and 112, have been described. Note that this can be done.

The closed contour generation unit 320 combines the position information from the buffer 318 with the contour information input through the line L10 to form a new closed contour 115 ′ in the bounding rectangle as shown in FIG. 6B. Form. In addition, the closed contour generation unit 320 includes the contour information on the new closed contour 115 ′, the index data of the new contour 115 ′, and the position and size information of the boundary rectangle included in the current VOP on the line L50. Add a new VOP to line L60
Is output to the optimum contour determination unit 350 through. The input to the optimum contour determining unit 350 is a new VOP having contour information for a new closed contour 115 ′ on the line L60, a current VOP on the line L12, or a plurality of VOPs on the line L30. Any one of the previously reconstructed contour video signals. The optimum contour determining unit 350 selects the closed contour information on the line L60 and the line L1 out of the predicted contours existing in a predetermined search area (for example, 8 pixels) of the contour included in the current VOP.
2, after detecting a predicted contour line most similar to any one of the contour line information included in the current VOP as the optimal contour line, the displacement between the contour line of the current VOP and the optimal contour line is represented. The motion vector MV and the index data of the optimal contour are output. The optimal contour determining unit 350 determines the optimal contour and the MV using a normal optimal contour determining method well-known in the art, and a description thereof will be omitted. According to a preferred embodiment of the present invention, the index data of the optimal contour has the same value as the label of the object pixel corresponding to the predicted contour. Thereafter, the index data of the MV and the optimal contour are supplied to the MUX 800 via the line L40 and to the motion compensator 400 via the line L20 as shown in FIG.

Referring back to FIG. 4, the motion compensator 400
Stores the optimal contour information on the frame memory 7 based on the MV on the line L20 and the index data of the optimal contour.
A predicted current contour is generated by extracting from 00 through line L30. The predicted current contour is an optimum contour shifted by MV. The motion compensating unit 400 supplies the predicted current contour information indicating the position data of the contour pixel of the predicted current contour and its index to the inter-coding unit 200 and the contour reconstruction unit 600 via the line L55.

The inter-coding unit 200 uses a normal vertex determination technique, based on the predicted current contour information on the line L55 and its index data, and the first and second contours of the current VOP on the line L10. Determine two vertices. Thereafter, the inter encoder 200 encodes the position data of the first and second vertices, and supplies the encoded vertex data to the MUX 800 and the inter decoder 500, respectively. MUX800
Multiplexes the encoded vertex data, the MV and the index data of the optimal contour, and sends the encoded contour data to a transmitter (not shown) for transmission.

On the other hand, the inter-decoding section 500 decodes the received encoded vertex data into decoded vertex data representing the decoded first and second vertex data and supplies the decoded vertex data to the contour reconstruction section 600. The contour reconstructing unit 600 stores the decoded vertex data from the inter decoding unit 500 and the line L5
5, and reconstructs the current VOP using the predicted current contour information input through step 5. Reconstructed current VO
P is stored in the frame memory 700 as a reconstructed front contour video signal for the next VOP.

While the preferred embodiment of the present invention has been described above, those skilled in the art will be able to make various modifications without departing from the scope of the present invention.

[0027]

Therefore, according to the present invention, an optimum contour is detected by detecting the edge of a predetermined boundary rectangle surrounding a VOP intersecting with the contour of the object and forming a closed contour. can do.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing positions of luminance data and chrominance data expressed in a 4: 2: 0 format.

FIG. 2 is a schematic diagram of a VOP expressed in a bounding rectangle.

FIG. 3 is a schematic diagram showing luminance shape information expressed in a binary mask form.

FIG. 4 is a schematic block diagram of an apparatus for encoding VOP data according to the present invention;

FIG. 5 is a detailed block diagram of a motion estimator in FIG. 4;

FIG. 6 is a schematic diagram illustrating a process of detecting the edge of a predetermined boundary rectangle surrounding a VOP that intersects with the contour of an object.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Boundary rectangle 20 Minimum rectangle 30 Object 100 Contour detection part 111-114 Each side of boundary rectangle 115 Contour 115 'Closed contour 200 Inter coding part 300 Motion estimation part 310 Intersection detection part 312 Start / end point determination part 314 Side Selection unit 316 Position information generation unit 318 Buffer 320 Closed contour generation unit 330 Control unit 332 Switch 350 Optimal contour detection unit 400 Motion compensation unit 500 Inter decoding unit 600 Contour reconstruction unit 700 Frame memory 800 MUX

Claims (10)

[Claims] 1. An edge detection method for detecting an edge of a predetermined boundary rectangle surrounding a video object plane (VOP) intersecting an outline of an object represented in a video signal, wherein the edge of the VOP is detected. Receiving VOP data having outline data representing the position of a pixel and information relating to the predetermined boundary rectangle, and detecting whether or not the outline of the VOP data intersects the edge of the predetermined boundary rectangle. A, determining an intersection on the edge of the predetermined boundary rectangle where the outline intersects, and based on the outline data, the intersection and the information on the boundary rectangle, Between the intersections along the corresponding side of the predetermined boundary rectangle corresponding to the intersection, based on the position information, sequentially obtain the position information of a plurality of points located along the side of the boundary rectangle between the intersections Be located Step c of generating an additional contour line corresponding to a portion composed of a plurality of points and generating additional contour line information for the corresponding additional contour line; and combining the contour line data and the additional contour line information. ,
Generating a closed contour within the predetermined boundary rectangle. D. 2. The method according to claim 1, wherein the number of intersections is two. 3. The edge detecting method according to claim 2, wherein the information of the boundary rectangle is information indicating a position and a size of the predetermined boundary rectangle. 4. The step (b) comprises, among the two intersections, an intersection which first appears counterclockwise from the inside to the outside of the predetermined boundary rectangle as a start point, and another intersection is determined as an end point. No. b1
And b. Selecting any one of the four sides of the boundary rectangle where the start point is located and selecting it as a target side
Two steps, and, when the end point is located on the target side, positions of a plurality of points located between the start point and the end point based on the outline data and the position and size information of the boundary rectangle. B3 step of generating information; and if the end point is not located on the target side, the start point and the target side are moved in a counterclockwise direction based on the contour data and the position and size information of the boundary rectangle. After generating position information of a plurality of points located between the point immediately before the last point of the boundary rectangle and selecting the next side in a counterclockwise direction among the four sides of the boundary rectangle, B4 step of selecting as the target side, and, if the end point is located on the new target side, the new point on the target side based on the outline data and the position and size information of the boundary rectangle. Step b5 of generating position information of a plurality of pixels located in a counterclockwise direction between the last point to be the start point of the target side and the end point, and the end point is not located on the new target side Step b4 and step b) until position information of all points located between the start point and the end point is obtained.
3. The edge detecting method according to claim 2, further comprising: b6 step of repeating 5 steps. 5. The step (b) sequentially stores position information of each point located between the start point and the end point in a buffer. All the position information is stored in the buffer, and the corresponding position information is stored in the buffer. The edge detection method according to claim 3, further comprising the step of: 6. An edge detecting device for detecting an edge of a predetermined boundary rectangle surrounding a video object plane (VOP) intersecting with an outline of an object represented in a video signal, wherein the VOP comprises:
Receiving the VOP data having the contour line data indicating the position of the contour line pixel and the information related to the predetermined boundary rectangle, and determining whether or not the contour line of the VOP data intersects the edge of the predetermined boundary rectangle. When detecting and intersecting, the contour line detecting means for determining an intersection on the edge of the predetermined boundary rectangle where the contour intersects, the contour data, the intersection, based on the boundary rectangle information, The position information of a plurality of points located along the side of the predetermined boundary rectangle between the intersections is sequentially obtained, and based on the position information, along the corresponding side of the predetermined boundary rectangle corresponding to the intersection. An additional contour line information generating means for obtaining an additional contour line corresponding to a portion consisting of a plurality of points located between the intersections and generating additional contour line information for the corresponding additional contour line; Combined data and the said additional contour information,
A closed contour generating means for generating a closed contour within the predetermined boundary rectangle. 7. The edge detecting device according to claim 6, wherein the number of the intersections is two. 8. The edge detection device according to claim 7, wherein the information of the boundary rectangle is information indicating a position and a size of the predetermined boundary rectangle. 9. The additional contour line information generating means determines, as a starting point, an intersection which first appears in a counterclockwise direction from the inside of the predetermined boundary rectangle toward the outside of the two intersections, and determines another intersection. A first arranging unit that negotiates as an end point; and a second side that selects any one of the four sides of the boundary rectangle where the start point is located and selects it as a target side.
Arranging means, when the end point is located on the target side, positions of a plurality of points located between the start point and the end point based on the outline data and the position and size information of the boundary rectangle. First position information generating means for generating information; and when the end point is not located on the target side, based on the contour data and the position and size information of the boundary rectangle, the start point and the counterclockwise direction. After generating position information of a plurality of points located between the last point on the target side and a point located immediately before the last point, the next side is selected in a counterclockwise direction from the four sides of the boundary rectangle. A second position information generating means for selecting as a new target side, when the end point is located on the new target side, based on the outline data and the position and size information of the boundary rectangle, The second position information generating means for generating position information of a plurality of pixels located in a counterclockwise direction between the last point to be the start point of the new target side and the end point on the target side, and the end point is If not located on the new target side, repeating means for repeatedly performing the operation of the second and third position information generating means until the position information of all points located between the start point and the end point is obtained; The edge detecting device according to claim 8, comprising: 10. The contour line information generating means sequentially stores position information of each point located between the start point and the end point in a buffer, and stores all position information in the buffer. The edge detection device according to claim 9, further comprising a storage unit that outputs position information.