Vehicle event analysis using path clues under occlusions

• Main Authors: Chia-Lung Lin, 林嘉隆
• Other Authors: Jun-Wei Hsieh
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/71316464898325443749

碩士 === 國立臺灣海洋大學 === 資訊工程學系 === 100 === The Background Subtraction method has extensively employed in movement detection for image process due to the merits of small amount of calculation and favorable detection result. However, if the method is used for detection under sophisticated environment it can not identify the variety of moving object. To solve this problem, the thesis has proposed the use of Speeded Up Robust Features、Shape Context for feature matching. Then, the thesis has proposed the use of the path feature for event analysis. First of all, path feature are extracted by detecting motorcycle from motorcycle tracking system techniques. Second, a symbol is composed of the features which obtain from path feature. Third, an event defined of classification by Support Vector Machine.