一种基于高斯模型的运动目标检测算法  

The Algorithm Of Moving Objects Detection Based on Gauss Mode

作　　者：孙南[1] 罗以宁[1] 柳伟[1] 

机构地区：[1]四川大学计算机学院

出　　处：《微计算机信息》2009年第31期186-188,共3页Control & Automation

摘　　要：针对传统分割算法易受噪声影响、分割完整性较差等缺陷,本文提出了一种基于高斯模型的运动目标检测算法。该算法在背景差分法的基础上,采用统计推断的方法获取分割阈值,检测运动目标,其中通过在估计的未变化背景区域中提取子块的方法实现阈值优化,增强了目标检测的鲁棒性。实验结果表明,该算法能够提高运动目标检测的准确性,具有较好的稳定性和实时性。In terms of some disadvantages of incomplete resisting noise and information for thresholding segmentation in the traditional algorithm, a moving objects detection algorithm based on Gauss model is proposed. The algorithm attained the threshold value of sil- houette and detected moving objects using statistical inference approach based on the background differencing, the method of extracting suhblock from the estimated unmoving part optimize threshold value, enhance the robust of object detection. Experimental results show that, the algorithm can improve accuracy of moving objects detection, it is more stability and has high real-time processing performance.

关 键 词：运动目标检测 高斯模型 图像分割 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]