Font Size: a A A

Complex Scene Target Detection Technology Research And Dsp System Realization

Posted on:2008-04-22Degree:MasterType:Thesis
Country:ChinaCandidate:W LiuFull Text:PDF
GTID:2208360212478910Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
In the field of computer vision and intelligent video surveillance, moving object detection is one of the fundamental tasks for video analysis, and it is very important. The subsequent technologies such as moving object classification, tracking and behavior understanding strongly depend on the results of moving object detection. In this thesis, we focus on the background modeling of complex environment and moving object detection, and a system for moving object detection based on TMS320DM642EVM is finished. The main contributions are as follows:1. Several background modeling algorithms are widely researched and analyzed based on extensive experiments, and the methods are compared each other in accuracy, speed, memory requirement and scenes that the algorithms can be applied.2. The algorithm based on kernel density estimation is analyzed. Moreover because of the big computation load, the low speed and the "dead lock" or false negative of the algorithm, a fast algorithm for moving object detection based on kernel density estimation and model switching is proposed. Experimental results demonstrate the good detection performance for real-time moving object detection and the low false positive of the algorithm.3. A detailed research about the hardware system is made, including the characteristics and the structure of the DSP chip, the function of the DSP EVM. Method of developing advanced video applied program in the developed environment CCS is analyzed and a research of the compositive tools of CCS is made.4. A detailed research about the develop process of the software system is made, including the principles of the system, the drive design, the definition of the data structure, the capture and the display of the video, the transition of the video format. At last a system for moving object detection based on TMS320DM642EVM is finished. Extensive experimental results show that the system has good performance for moving object detection, relatively fast processing speed, good stability and expansibility and can be used flexibly and neatly.
Keywords/Search Tags:Background Modeling, Moving Object Detection, DSP, Kernel Density Estimation
PDF Full Text Request
Related items