Instructor Dr K R Rao Presented by Rajesh Radhakrishnan Not much work had been presented in evaluating the functionality of the compressed domain object detection with that of the spatio temporal one More research work is going on to improve the efficiency of compressed domain object detection to be used for computer vision application The key parameter to perform object detection is to determine the optical flow in case of spatialtemporal detection and motion vector estimate in case of compressed domain detection Fig1 4 Block diagram of spatio temporal object detection Fig 2 eg frame 8 an input to explain moving object detection Following code were generated in MATLAB Raw image obtained after frame differencing Fig 3 Background modeling by frame differencing Fig 1 Fig 2 Fig 4 Foreground detection using threshold model 10 Fig 4 1 threshold model 40 Fig 4 2 Fig 5 Data validation This is a parametric model of skin detection Only motion vector information is required Recent work in compressed domain object detection is by vector featured image algorithm 2 This algorithm is efficient enough to detect pauses in moving object Initial region extraction Involves converting data from encoded format to display format 1 Form a initial region definition making use of current block Bc reference block Br and background block Bb 2 2 Moving region detection involves labeling of blocks as moved Bm and unmoved region bu These five blocks form the basis of object detection algorithm 3 Modification of vector featured regions This is the module where the pauses in moving object are detected 4 Final step involves moving object tracking Fig 6 Initial region extraction 2 Three Blocks directly extracted from motion vectors they are current block Bc reference block Br and background block Bb Fig 7 updating of moving and unmoving regions 2 Here additional two blocks are included to detect object stops they are moving block Bm and unmoving block Bu A mapping is done between the current and the next frame and new regions are marked by Bc and overlapping regions as Bm and non zero to zero motion vector are marked as Bu To get moving object regions extract minimum bounding rectangles MBR s which mark the regions of moving object in a video To implement a compressed domain object detector that can recognize moving hand location To implement a compressed domain based moving object detector Then generate a time series containing a centroid of detected object with detection box size of 40x40 Three essential modules are required to obtain the comparative study of the moving object detection between two domains First is the manual annotation of hand locations using a GUI to get the co ordinate location of the hand in every frame Second is to obtain time series of hand locations based on spatio temporal algorithm Third is to obtain time series of hand locations based on compressed domain algorithm Hand locations of the detected hand are going to be compared with annotated hand locations to find efficiency Efficiency to detect multiple hand locations and execution time of both the algorithms will be tested More parameters may be added for future considerations Z Qiya and L Zhicheng Moving object detection algorithm for H 264 AVC compressed video stream ISECS International Colloquium on Computing Communication control and management pp 186 189 Sep 2009 T Yokoyama T Iwasaki and T Watanabe Motion vector based moving object detection and tracking in the MPEG Compressed Domain Seventh International Workshop on content based Multimedia Indexing pp 201 206 Aug 2009 Kapotas K and A N Skodras Moving object detection in the H 264 compressed domain International Conference on Imaging systems and techniques pp 325 328 Aug 2010 Sen Ching S C and C Kamath Robust techniques for background subtraction in urban traffic video Center for Applied Scientific Computing Lawrence Livermore National Laboratory Jul 2004 S Y Elhabian K M El Sayed Moving object detection in spatial domain using background removal techniques state of the art Recent patents on computer science Vol 1 pp 32 54 Apr 2008 O Sukmarg and K R Rao Fast object detection and segmentation in MPEG compressed domain TENCON 2000 proceedings pp 364 368 Mar 2000 W B Thompson and Ting Chuen P Detecting moving objects International journal of computer vision pp 39 57 Jun 1990 Thank You
View Full Document
Unlocking...