基于条件随机场的视频运动目标可靠性检测方法  

Reliability Detection Method of Video Moving Target Based on Conditional Random Field

作　　者：李晓峰[1] 邢金明[2] LI Xiao-feng;XING Jin-ming(Department of Information Engineering,Heilongjiang International University,Harbin 150025,China;Northeast Normal University,Changchun 130024,China)

机构地区：[1]黑龙江外国语学院信息工程系,黑龙江哈尔滨150025 [2]东北师范大学,吉林长春130024

出　　处：《计算机技术与发展》2020年第7期60-65,共6页Computer Technology and Development

基　　金：教育部科技发展中心产学研创新基金(2018A01002);教育部人文社科基金项目(15YJC890004);中国博士后基金项目(2017M610852);吉林省社科基金重点项目(2016A5)。

摘　　要：针对当前视频运动目标检测方法受到环境随机场因素的影响,存在检测性能不佳的问题,主要由视频运动特征分布散乱,跟踪效果差所导致,提出基于条件随机场模型的视频运动目标可靠性检测方法。首先,采用模板匹配方法构建视频运动目标的像素特征点块匹配结构模型,获取目标统计特征量,划分统计特征量中的关键帧频带;其次,根据关键帧频带划分结果,构建帧内编码函数,跟踪目标运动轨迹;再次,依据跟踪结果构建目标图像的条件随机场分布模型;最后,依据运动目标的边缘轮廓特征分布,对目标进行粗重构,并增加图像特征,实现目标的高分辨视觉重建,以此完成视频运动目标可靠性检测。实验结果表明,采用该方法进行视频运动目标检测的检测准确率、跟踪精度均较高,且检测时间较少,检测分辨率较高,具有一定的可靠性。Aiming at the problem that the current video moving target detection methods are affected by the environmental random field factors and have poor detection performance,which is mainly caused by the scattered distribution of video moving features and poor tracking effect,we propose a video moving target reliability detection method based on conditional random field model. Firstly,the pixel feature block matching structure model of video moving target is constructed by the template matching method to obtain the statistical feature of the target and divide the key frame frequency band of the statistical feature. Secondly,according to the result of the key frame frequency band division,the intra-frame coding function is constructed to track the target motion track. Thirdly,according to the tracking knot,the conditional random field distribution model of the target image is constructed. Finally,according to the edge contour feature distribution of the moving target,the target is roughly reconstructed,and the image features are added to realize the high-resolution visual reconstruction of the target,so as to complete the reliability detection of the moving target in video. The experiment shows that the proposed method has higher detection accuracy and tracking accuracy,less detection time,higher detection resolution and certain reliability.

关 键 词：条件随机场模型 视频 运动目标 可靠性 检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]