Arnaud Eric Jacquin

Motion-adaptive modelling of scene content for very low bit rate model-assisted coding of video sequences

A method and apparatus for video coding whereby a region of an image which includes a predetermined object such as a persons face in the foreground portion of the image is automatically determined. Specifically, the foreground portion of the image is identified, and one or more predetermined (geometric) shapes (e.g., ellipses) are compared with the shapes of objects found in the foreground portion of the image. The foreground portion of an image may be determined by performing a global motion estimation of the overall image to detect global image movement resulting, for example, from camera pan and zoom. That portion of the image whose movement is consistent with the estimated global motion may be identified as the background portion, with the remainder of the image identified as the foreground portion. The identified region of the image which includes the predetermined object and portions of the image which do not include the predetermined object may be coded with differing levels of coding accuracy (e.g., using different quantization levels), such that if the identified region contains, for example, a persons face, the quality of the coding of the face may be improved relative to the quality of the coding of other portions of the image.

A normalized rate-distortion model for H.263-compatible codecs and its application to quantizer selection

This paper proposes a new normalized parametric rate-distortion model which can be efficiently applied to H.263-compatible video codecs and other codecs of similar type. The model is very efficient in terms of computational complexity and amount of required memory since it approximates all 31 rate-distortion relations (one for each quantization parameter QP) with only one curve by normalizing the energy of signals and the quantization distortion with functions of QP. We show how the model parameters can be obtained from experimental data. We also propose a quantizer selection scheme which uses the model to estimate the number of bits required for encoding 8/spl times/8 residual blocks. Experiments show that the model provides very good approximations of the rate and the distortion characteristics for all QPs, and that the proposed quantizer selection scheme works better than the scheme used in the MPEG-4 Video Verification Model Version 5.1.

A real-time scalable software video codec for collaborative applications over packet networks

Layered coding architectures are attractive in theory for two reasons. FirsL they natur~y allow for heterogeneity in networks and receivers in terms of chent processing capabfity and network bandwidth. Second, they correspond to optim~ ut~iation of avfiable bandwidth when seved video qutity Ievek are desired. b practice, the increased complexity imposed by the Iayering constraint can force the designer to do away with key parts of a coding algorithm, e.g. motion estimation for video coding. h this paper, we propose a salable sofware-ody video codec architecture with motion est”hnation, which is suitable for r~-tirne audio and video communication over packet network. Thecodingdgorithrnis compatiblewith~T recommendation H.263+ and includes various techniques to reduce compIefity. Fast motionestiation @~) is performed at the H.263compatibIe base layer and used at higher Iayers, and perceptual macroblock skipping is performed at dl layers before h~. SNR sdabfity

Constrained wavelet packets for tree-structured video coding algorithms

Traditional wavelet packet (WP) optimization techniques neglect information about the structure of the lossy part of the compression scheme. Such information, however, can help guide the optimization procedure so as to result in efficient WP structures. We propose a wavelet packet algorithm with a constrained rate-distortion optimization which makes it suited to subsequent tree-structured coding such as with the set partitioning in hierarchical trees (SPMT) algorithm. The (octave-band) wavelet transform lends itself to simple and coherent tree-shaped spatial relations which can then be used to define zero-trees. Yet, input images have different frequency distributions and an adaptive transform such as WP is bound to be more efficient on an image-by-image basis. With WP algorithms, the coefficients in the WP domain can be rearranged to produce what resembles (or simulates) the normal wavelet transform structure. This stage is usually performed to simplify the coding stage. However, an unconstrained optimization can result in a transformed image with complicated or incoherent tree-shaped spatial relations. This work aims to show that the efficiency of embedded coders such as SPMT and Shapiros Zerotrees strongly depends on WP structures with coherent spatial tree relationships.