基于DSP的无背景模型车流量检测

针对车流量检测中遇到光照变化、树阴、树枝摇动等一系列问题,在运动物体法,背景建模法和时间差分法的基础上,提出了一种基于DSP的无背景模型的新型车流量检测方法。笔者依次使用基于块的帧差法、双重前景融合法、基于纹理对象分割法、假前景滤波法对车流量进行检测。经实验验证,采用无背景模型算法能成功有效地检测车流量,检测准确率达88%,提高了车流量检测的精度。

The No-background Model for Flow Detection Based on DSP

A series of problems were encountered in the flow detection,such as illumination variation,shad- ows,and branches shaking.On the bases of the method of moving objects,background modeling method and the time difference method.They propose a new no-background model for flow detection based on DSP.In turn use the block-based frame difference method,the dual foreground of fusion method,segmenta- tion method based on texture object,false foreground filtering method to detect traffic flow.The experiments prove that no background model algorithm can effectively detect traffic flow detection.The accuracy rate reach 88%,the precision of flow detection is better improved.