局部熵差图像帧间动态位移信息并行滤波  

Parallel Filter Simulation of Dynamic Displacement Information Between Local Entropy Difference Images

作　　者：黄伟建[1] 王月兴 黄远[1] HUANG Wei-jian;WANG Yue-xing;HUANG Yuan(School of Information&Electrical Engineering,Hebei University of Engineering,Handan Hebei 056000,China)

机构地区：[1]河北工程大学信息与电气工程学院,河北邯郸056000

出　　处：《计算机仿真》2020年第7期296-300,共5页Computer Simulation

基　　金：河北省自然科学基金资助项目(F2015402077);河北省高等学校科学技术研究项目(QN2018073)。

摘　　要：针对当前滤波方法存在滤波后图像视觉效果不佳,信噪比和保真度不高等问题,提出了基于局部熵差的图像帧间动态位移信息并行滤波方法。在运动目标图像粗分割基础上,将运动区域和非运动区域分离,并采用背景差分法对分离出来的运动区域做进一步分割,实现背景图像中的运动目标检测。为了提高后续目标跟踪和检测准确率,在HSV颜色空间检测运动目标区域的阴影部分。在此基础上,将图像帧的熵值作为运动目标的一个全局特性,每个类别的中心帧被选取为运动目标区域一个镜头的关键帧,将提取获得的关键帧中的一个与其它剩余关键帧做比较,利用局部熵差法计算图像帧熵值标准偏离度,消除冗余帧影响,实现图像帧间动态位移信息并行滤波。仿真结果表明,提出方法滤波后图像信噪比高、保真度高,且具有较好的并行性能。Due to poor visual effect,low signal-to-noise ratio and low fidelity of current filtering methods,this article puts forward a parallel filtering method of Inter-frame dynamic displacement information based on local entropy difference.On the basis of rough segmentation of motion object image,motion region and non-motion region were separated.Meanwhile,the separated moving region was further segmented by background difference method,and thus to realize the detection for motion object in background image.In order to improve the accuracy of following target tracking and detection,the shadow part of motion object region was detected in HSV color space.On this basis,the entropy of image frame was regarded as a global character of motion object.In addition,the central frame of each category was chosen as the key frame of a scene in motion target area.After the extraction,a key frame was compared with other remaining key frames.Finally,the local entropy difference method was used to calculate the standard deviation of image frame entropy value,and thus to eliminate the influence of redundant frames.Thus,the parallel filtering of inter-frame dynamic displacement information was achieved.Simulation results show that the proposed method has higher signal-to-noise ratio,higher fidelity and better parallel performance.

关 键 词：局部熵差 图像 帧间动态位移信息 并行滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]