Implementation on video object segmentation algorithm Kuo, Yi-Ting and Wu, Chia-Peng May 03. 2004
Outlines Introduction Algorithm Architecture of hardware implementation Systolic array for texture feature extraction
Introduction Our project is focused on extracting moving objects from video. The algorithm of moving object segmentation can be applied to MPEG-4 standard which enable content-based functionality. Also can be used in traffic surveillance system.
Algorithm Change detection Previous frame In-1 Current Frame In Find moving object edge Smooth edge Moving object -
Mean and Variance Features The two features (mean and variance),ft1(m,n) and ,ft2(m,n) are textural appearance of the area surrounding a pixel (m,n) in a small window centered on this pixel, Nw is the number of pixel of Ws * Ws of window W .
Systolic array for texture feature extraction
Systolic array for extracting the two texture features ft1, ft2 Systolic array for extracting the two texture features using 5x5 window The luminance component of a reference frame fy(m,n) are scanned Into 1+4Nc size FIFO. Nc = number of columns of reference frame Block A: accumulates luminance components. Block M: generate a mean value by dividing the accumulated result by Nw Block V: calculate localvariance texture feature.
DEMO
References [1] Changick Kim and Jenq-Neng Hwang, “Fast and automatic video object segmentation and tracking for content-based application,” IEEE Trans. Circuits and Systems for Video Technology, vol. 12, No. 2, Feb. 2002, pp. 122-129. [2] J. F. Canny, “A computational approach to edge detection,” IEEE Trans. Pattern Anal. Machine Intell., vol. PAMI-6, pp. 679-698, Nov. 1996. [3] Jinsang Kim and Tom Chen, “Real-time video objects segmentation using a highly pipelined microarchitecture,” Proceedings of the IASTED International Conference, Visualization, Imaging, and Image Processing, Sep. 3-5, 2001, Marbella, Spain, pp. 483-488 [4] Rafael C. Gonzalez and Richard E. Woods, Digital Image Processing.