基于三帧差分时空熵的运动目标检测算法  

The Moving Object Detection Based on Three Image Difference and Spatio-Temporal Entropy

作　　者：张晓丽[1] 

机构地区：[1]西安航空学院计算机学院,陕西西安710077

出　　处：《自动化技术与应用》2017年第7期78-82,共5页Techniques of Automation and Applications

基　　金：陕西省网络计算与安全技术重点实验室项目(编号15JS 078);西安市科技计划资助项目(编号CXY1518(1))

摘　　要：图像序列中的运动目标检测是一个相当重要又非常困难的问题。针对这一问题,提出了一种基于三帧差分时空熵的运动目标检测算法。算法首先使用三帧差分的方法计算差分图像,然后通过计算差分图像中像素的时空熵并进行形态学滤波和阈值分割,可以有效的检测出运动目标。实验结果表明,该方法运算复杂度低,抗噪能力较强,易于实现实时处理。The moving object detection is a very important and very difficult problem in image sequence. This paper proposes a method of motion detection based on three frame difference and spatio-temporal entropy. The method first calculates the differential image by three-frame difference method, and then calculates spatio-temporal entropy of the pixel in differential image. After morphological filtering and thresholding, the moving object can be detected effectively. Experimental results show that this method has low computational complexity, anti-noise ability, easy to implement in real-time.

关 键 词：三帧差分 时空熵 运动目标检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]