Rate-constrained Change detection and Shape Coding

		Xinqiao (Chiao) Liu
	Information Systems Laboratory
	Dept. of Electrical Engineering

	    chiao@stanford.edu

MPEG-4 is the first international standard allowing the transmission
of arbitrarily shaped video objects(VO's). Several schemes have been
proposed in the object-based video coder to encoding the VO shape
information, such as binary bitmap-based shape coder and binary
contour based shape coder[1]. A polygon based shape
representation was developed for OBASC (object-based
analysis-synthesis coder)[2]. Under the constrain of the maximum
distortion D_max for a lossy shape encoding, the algorithm searches
for a polygon approximation which requires the minimum number of
polygon points (vertex). Splines that are defined by the vertices may
also be used in the last step to smooth the representation of the
shape. Vertex coordinates and the curve type between two vertices are
arithmetically encoded. The minimum number of vertices used in the
polygon approximation of the original object shape results in the best
bit-rate achievable.

Motion segmentation is performed in the image analysis unit of the
OBASC. The efficiency of the shape coding as a result of motion
segmentation scheme, however, is not discussed. A new motion
segmentation technique is proposed based on a new design technique of
the video image sensor. The contour of a moving object is generated by
performing motion detection inside each light sensing pixel during
frame exposure time. This scheme enables the study of the correlation 
between change detection and shape coding using original input to the
video sensor.

In this project, polygon-based lossy contour shape encoder will be
implemented first. The correlation between the shape encoder and the
proposed motion segmentation scheme will then be studied where the
distortion function is redefined as the difference between polygon
approximated shape and the original moving object. Rate constrained
change detection and shape coding will be optimized.

[1] Katsaggelos, A.K.; Kondi, L.P.; Meier, F.W.; Ostermann, J.;
Schuster, G.M. "MPEG-4 and rate-distortion-based shape-coding
techniques", Proceedings of the IEEE, Volume: 86 6 , June 1998 ,
Page(s): 1126 -1154

[2] Hotter, M. "Optimization and efficiency of an object-oriented
analysis-synthesis coder", IEEE Transactions on Circuits and Systems
for Video Technology, Volume: 4 2 , April 1994 , Page(s): 181 -194