一种改进的帧间差分与背景差分融合的运动车辆检测方法

An Improved Moving Vehicle Detection Method Based on Integration of Interframe Difference and Background Difference

针对运动车辆检测算法当前所存在的环境适应性差、检测信息不完整的问题,在帧间差分法与边缘检测法的基础上,提出了一种改进的帧间差分运动车辆检测算法,以解决帧间差分法检测车辆轮廓不完整的问题。而后将改进帧间差分法融合混合高斯背景模型得到最后的结果,以解决采用单一的帧间差分法检测到的运动车辆内部信息不完整和采用单一混合高斯背景模型对光照敏感的问题。实验结果表明,所提出的方法提高了对车辆检测的准确性以及对外部环境变化的适应能力。

Aiming at the problems with existing moving vehicle detection algorithms such as poor environmental adaptability andincomplete detection information, this paper proposes an improved interframe difference moving vehicle detection algorithm based on theinterframe difference method and the edge detection method, in a bid to improve the interframe difference method which tends to come outwith incomplete vehicle contours. Then the improved interframe difference method is integrated with the Gaussian mixture background modelto obtain the final result, so as to prevent the moving vehicle internal information fragmentation prone to occur with the single interframedifference method and the over-sensitivity to illumination with the single Gaussian mixture background model. At the end, the experimentalresult shows that the method proposed can improve the accuracy of vehicle detection and exhibits higher adaptability to changes in theexternal environment.