Video Object Extraction Integrating Temporal-Spatial Information

A novel video object segmentation method combines change detection and edge detection is proposed here. The process of the algorithm can be divided into three parts: motion detection, spatial segmentation and temporal-spatial filter, which integrates the spatial and temporal information of the video sequence. The motion detection step makes use of the t -distribution significance test; then, the initial motion detection mask of every frame in the video sequence can be integrated to form the movement mask. For spatial segmentation the Sobel edge detection operator is used to get the boundary of the video objects in current frame; temporal-spatial filter then integrates the temporal and spatial information, extracts the precise boundary of moving object and also reduces the residual noise. Finally, the segmentation of video objects can be optimized by filling and morphology operation.