Motion detection in dynamic scenes based on fuzzy c-means clustering
Motion detection in dynamic scenes is an extensively applicable technology but a difficult subject in computer vision. It is less developed compared with motion detection in static scenes. In the paper we present a novel approach to detect moving object in dynamic scenes without any prior information about moving object or dynamic scenes. It is mainly based on the fact that the displacements of features from moving object and background are different evidently, which can be used as a criterion for distinguish the moving object and background. SIFT (Scale Invariant Feature Transform) algorithm is used to detect feature points, an initial match set is obtained by Euclidean metric and the rule of nearest neighbor distance ratio, and a consistency test is performed to obtain robust feature correspondences. With the displacement vectors generated from the robust feature correspondences, a fuzzy c-means clustering algorithm is used and the feature points from moving object are detected accurately. The effectiveness of the proposed method is demonstrated using real video sequences from moving cameras.
motion detection dynamic scenes scale invariant feature transform fuzzy c-means clustering
Xiaqiong Yu Xiangning Chen Mengnan Gao
Academy of Equipment Command and TechnologyBeijing, 101416, China Wuhan Ordnance NCO Academy of PLA Wuhan, 430075, China
国际会议
哈尔滨
英文
191-195
2011-01-18（万方平台首次上网日期，不代表论文的发表时间）