基于二重差分法的光流场运动检测

对视频图像序列中运动目标的检测,是计算机视觉领域长期关注的问题.传统的Horn-Schunck光流算法非常耗时,不利于在视频图像序列中做实时分析,故提出采用结合背景差分法与帧间差分法的二重差分法(DS)来做光流算法的预处理.实验结果也表明二重差分法的光流算法(DSOF)比单纯用Horn-Schunck方法计算光流效率更高,用时更短,能更好地满足对视频图像序列做实时运动检测分析的需求.

Optical Flow Calculation Based on Dual Subtraction for Motion Detection

Detecting motion object in sequence frames is an important subject in Computer Vision field. Horn-Schunck algorithm is normally used for that, but it is not suitable for real-time analysis of sequence frames, because it will cost too much time on calculating. As a result, a novel motion detection algorithm called DSOF(Optical Flow Calculation Based on Dual Subtraction) was proposed in this paper to solve the real-time problem, which combines background subtraction with frame subtraction to prepare the ground for optical flow calculation. The result of experi-ment demonstrates that the novel algorithm is more effective, rims more quickly, and it would satisfy the demands of real-time analysis in sequence frames.