Motion object segmentation based on spatio-temporal combined and dual-restricted Snake algorithm
基于时空联合双重约束Snake算法的运动目标分割

A Spatio-Temporal Combined(STC) Snake algorithm for segmenting motion objects was presented.Firstly,by using the gray-level gradients of the pre-segmenting image and the temporal information of two adjacent frames,an external energy function with STC dual-restricted was constructed,which led the Snake curve to deforming and converging.Secondly,when initializing the Snake curve,by subtracting two adjacent frames,the approximate region of motion object was extracted.The length and width of the region's external rectangle were used as the axis's length to build an ellipse.Finally,N equal interval points were selected from the ellipse as the initial contour of Snake for later iterative computing.Experiment results show that the method is feasible and good,which can segment non-rigid motion objects accurately.