Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

A Research to the Algorithms of Moving Object Detection, Shadow and Ghost Detection and Removal

Author WuLiang
Tutor ZhouDongXiang
School National University of Defense Science and Technology
Course Information and Communication Engineering
Keywords Intelligent Visual Surveillance Moving Object Detection Mixed Gaussian Model Shadow Detection Ghost Detection abandoned objects detection
CLC TP391.41
Type Master's thesis
Year 2007
Downloads 369
Quotes 3
Download Dissertation

On the analysis of the current situation and development of the intelligent video surveillance system, a research was made for the key technology of the segmentation of moving objects and the detection and removal of both shadows and ghosts.The moving objects detection is the basis of tracking, classification, recognition and so on. After discussing various object detection algorithms elaborately, this paper uses a standard of qualitative estimate proposed in this paper to make a comparative analysis for the result of algorithms which were used widely for object detection; finally the MGM was adopted to detect moving objects.In order to remove the shadow of object foreground, a method of shadow detection based on the Phong lighting and radiosity model and post-processing for shadow removal is proposed on the basis of the summary and analysis for various shadow detection. Furthermore, a qualitative and quantitative estimate standard is used to make a comparative analysis for the result of previous and new algorithms. The experiment results prove that the new method is robust and offers more advantage over other algorithms presented previously on detection rate and discrimination rate.In order to resolve the problem of ghosts in moving objects detection based on background subtraction, a method of ghosts detection based on histogram matching was proposed and also used to detect abandoned objects after having analyzed pixels distribution of border area of every foreground blob. Experimental results show that the developed algorithm not only detects and removes ghosts fast and feasibly, and also overcomes the shortcomings of other algorithms.

Related Dissertations
More Dissertations