Image analysis – Image compression or coding – Shape – icon – or feature-based compression
Reexamination Certificate
2000-05-19
2003-02-04
Chen, Wenpeng (Department: 2624)
Image analysis
Image compression or coding
Shape, icon, or feature-based compression
C375S240080, C345S419000
Reexamination Certificate
active
06516093
ABSTRACT:
The present invention relates to a method of coding a sequence of pictures thanks to a segmentation operation of each of said pictures followed by a coding operation of each of the obtained successive partitions, and to corresponding coding and decoding systems. This invention is particularly useful in relation with the MPEG4 standard which will play within some years a key role in the field of all multimedia applications.
In the field of very low bit rate video coding, there is an increasing interest in region-based compression methods describing each image of a sequence in terms of a set of regions (called a partition) and of some information for each region to be used by the receiver to reconstruct the image. The main difference between these techniques relates to the importance they assign to the spatial or the motion information. For instance, the article “Object-oriented analysis-synthesis coding of moving images”, H. G. Musmann, M. Hotter and J. Ostermann, Signal Processing: Image Communication, vol.1, n°2, October 1989, pp.117-138, describes a coding scheme where motion plays the central role and the image is restored on the receiver side by motion compensation of past restored frames. A partition of each image is used to define the regions that should be compensated. This approach leads to good results if the sequence can actually be compensated, that is, if no new objects are allowed to be introduced in the scene and if scene changes are prohibited. As a result, this technique is mainly dedicated to very specific applications such as “head and shoulders” sequences.
Other approaches, more general, mainly deal with the spatial information of the scene. Coding algorithms combining a spatial analysis of the image sequence with a motion compensation of the transmitted information can also be proposed. On the one hand side, the spatial analysis is used to get a general scheme able to deal with any kind of sequences and scene changes. On the other hand side, motion information is used to increase the coding efficiency by compensation of the spatial information that has to be transmitted (partition and texture or color). An analysis of the results of such coding algorithms shows that about 55% of the bitstream is devoted to the partition information, about 35% to the texture (or color) and the remaining 10% to motion.
It is a first object of the invention to propose an improved coding method where much more texture information is sent to the receiver.
To this end the invention relates to a coding method such as indicated in the preamble of the description and wherein, for each current partition considered with respect to the previous one, said method comprises:
(1) an initialization step, by subdivision of said current picture into a mesh of blocks the nodes of which are located on high gradient points and of the blocks of said mesh into two triangles;
(2) a projection step, allowing to follow the time evolution of the sequence thanks to a motion estimation operation, for associating displacement vectors to the nodes of each triangle, followed by a mesh definition operation, for defining a new mesh on the basis of this motion of the nodes;
(3) a mesh motion coding step, in order to create the next partition and to define the motion field allowing to compensate the texture information, said compensation allowing to carry out a synthesis step of a prediction of the current picture, based on the knowledge of the nodes motion;
(4) a segmentation and merging step, for an adaptation of the topology of the mesh to the evolution of the scene, said segmentation sub-step allowing to introduce new triangles in the mesh and said merging sub-step allowing to remove small or degenerated triangles;
(5) a mesh update coding step;
(6) a texture error detection and coding step.
According to this method, it is proposed to work not with arbitrary shaped regions as in some previous embodiments but with simple shapes such as triangles and to send at the same time the motion and the partition information. This idea leads to the definition of an active triangular mesh coding scheme. Active meshes, studied for example in the communication “Active mesh: a feature seeking and tracking image sequence representation scheme”, IEEE Transactions on Image Processing, vol.3, n°5, September 1994, pp.610-624, are generally used to model the motion information but very few complete active mesh coding schemes have been proposed. In the present description, the mesh is used to model the sequence itself and plays two fundamental roles: (a) it defines a simplified partition made of triangles (the partition is used to define homogeneous regions in terms of texture and, because of its particular structure, the partition itself can be very efficiently coded), (b) the nodes of the triangular mesh are also used to send the motion information. This means that the motion of each region is assumed to be modelled by an affine transformation and can take into account rather complex events.
It is another object of the invention to propose a corresponding coding system, for carrying out said method.
To this end the invention relates to a system for coding a sequence of segmented pictures defining successive partitions, each current picture being originally available in the form of a mesh of blocks the nodes of which are located on high gradient points and each block being subdivided into two triangles, wherein said system comprises:
(A) a projection circuit, for estimating on the basis of the mesh defined for a previous picture (T-
1
) a projected mesh made of polygons and corresponding to the current picture T;
(B) a mesh coding circuit, for coding motion and texture associated to said projected mesh;
(C) a texture error detection and coding circuit.
According to a first embodiment, said mesh coding circuit may comprise:
(a) a mesh motion coding circuit, for coding a translation vector defined for each node of the mesh;
(b) a synthesis circuit, for a prediction of the current picture based on the knowledge of the nodes motion field and a subsequent compensation of the texture information;
(c) a mesh updating circuit, for a modification of the topology of the mesh by segmentation or merging of triangles;
(d) a mesh update coding circuit, for coding the modified information related to the topology of the mesh.
According to another embodiment, said mesh coding circuit may alternatively comprise:
(a) a mesh tree definition device, for associating to said projected mesh coarser meshes obtained by merging and which define upper levels of said tree and finer meshes obtained by resegmentation and which define lower levels of said tree;
(b) a decision sub-assembly, for taking the mesh proposals of said defined mesh tree and making a decision on which polygons will belong to a so-called final projected mesh and which coding technique among a predetermined list will be used within each one of said polygons;
(c) a coding sub-assembly, for coding said decision, said final projected mesh and the texture of each of said finally selected polygons.
In the case of this second embodiment, decision sub-assembly may then comprise a decision tree construction device, said decision tree being provided for conveying information about cost and quality of said coding techniques, and an optimization device, for finding a set of polygons defining said final projected partition with the highest coding quality and the lowest associated cost.
It is still another object of the invention to propose a system for decoding signals coded by means of a coding system such as the above-described one.
To this end the invention relates to a system for decoding signals coded by means of a coding system such as described above, said coded signals constituting for each current partition a coded information corresponding to an optimal partition composed either of regions of a main partition determined by a motion estimation and compensation of a previous partition and a temporal extension of the compensated partition or of regions of additional partitions created by mer
Ayuso Francisco X.
Marti Eva
Pardas Montse
Salembier Philippe
Chen Wenpeng
Gross Russell
Koninklijke Philips Electronics , N.V.
LandOfFree
Segmented video coding and decoding method and system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Segmented video coding and decoding method and system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Segmented video coding and decoding method and system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3118141