基于四帧帧差和混合高斯模型的运动目标检测

鉴于传统的帧差法检测准确率不高,容易造成检测错误等问题,提出了一种改进的视频序列运动目标检测算法.该算法将混合高斯模型与改进的四帧差分算法相结合:首先,改进的四帧差分是取连续的四帧——第1帧与第3帧、第2帧与第4帧分别进行差分二值运算,采用动态阈值以适应光线变化,将差分的结果轮廓填充,进行"与"运算;然后,将混合高斯建模后得到的运动目标与改进的四帧差分算法得到的运动目标,进行逻辑"与"运算;最后,通过形态学处理检测出运动目标.实验结果证明,改进的算法既能适应光照的变化,又能有效克服空洞的现象,与同类算法相比,具有更高的鲁棒性和准确率.

Moving Object Detection Based on Four-Frame Difference and Gaussian Mixture Model

In view of the low accuracy of traditional frame difference detection and its easiness to cause detection errors and other issues,this paper proposes an improved video sequence motion target detection algorithm.The algorithm combines the mixed Gaussian model with the improved four-frame frame difference algorithm.Firstly,the improved four-frame difference is a continuous four-frame,the first and the third,the second and the fourth are the difference.The dynamic threshold is used to fit the light change,and then the difference result contour is filled,and finally the " AND" operation is carried out.Do logical " AND" operation between the moving target obtained by mixing the Gaussian model and the moving target obtained by the improved four-frame difference algorithm,and finally through the morphological processing to detect the moving target.From the experimental results,it is proved that the improved algorithm can adapt to the change of illumination and overcome the phenomenon of void effectively,and it has higher robustness and accuracy than the similar algorithm.