Compressed domain object tracking for automatic indexing of objects in MPEG home video

R. Achanta1, M. Kankanhalli1, P. Mulhem2
1School of Computing, National University of Singapore, Singapore
2IPAL-CNRS, National University of Singapore, Singapore

Tóm tắt

Object tracking is of utmost importance for automatic indexing of video content. This work presents an object tracker that operates directly on MPEG compressed data. Motion vectors and discrete cosine transform (DCT) coefficients directly available from the compressed video stream are exploited for the purpose of tracking. Tracking proceeds in two steps: motion vector based tracking in P and B frames within the groups of pictures (GOPs), and object identification in I frames. Colour, which is one of the strongest cues for tracking is used for the identification step. Such a system offers speed, simplicity and robustness against occlusion and camera motion, with good intra-shot tracking for shots in excess of 500 frames, as shown in the experimental results.

Từ khóa

#Machine assisted indexing #Transform coding #Video compression #Tracking #Discrete cosine transforms #Cameras #Image segmentation #Home computing #Streaming media #Robustness

Tài liệu tham khảo

10.1007/3-540-45453-5_145 won, 1997, Extracting image features from MPEG-2 compressed stream, SPIE, 3312, 426 lee, 2001, A Real-time Colorbased Object Tracking Robust to Irregular Illumination Variation IEEE shen, 1996, Direct Feature Extraction from Compressed Images SPIE, 2670 achanta, 2002, Object Tracking in Compressed Domain MPEG Video 10.1109/93.839313 10.1109/76.836288 10.1109/ISCAS.1998.698810 lie, 2001, Tracking moving objects in MPEG-compressed videos, IEEE Int Conf Multimedia Expo chang, 1993, Compositing and Manipulation of Video Signals for Multimedia Network Video Services (Parts in 10.1117/12.298412 nakajima, 1999, Moving object detection and identification from MPEG coded data, International Conf on Image Processing