基于改进Renyi熵的运动目标图像提取仿真  

Moving Target Image Extraction Simulation Based on Improved Renyi Entropy

作　　者：王超[1] 宋路[2] 李长明 WANG Chao;SONG Lu;Li Chang-ming(School of Electrical Information Engineering,Changchun Guanghua University,Changchun Jilin 130033;School of Electronic Information Engineering,Changchun University of Science and Technology,Changchun Jilin 130022,China)

机构地区：[1]长春光华学院电气信息学院,吉林长春130033 [2]长春理工大学电子信息工程学院,吉林长春130022

出　　处：《计算机仿真》2020年第4期389-392,共4页Computer Simulation

基　　金：吉林省教育厅“十三五”科学技术研究规划项目(JJKH20181376KJ)。

摘　　要：运动目标图像提取有助于通过计算机视觉实现复杂变化背景下的图像细节分析,现有图像提取方法通常针对某一类图像作为研究对象,致使应用场合受到一定限制。针对运动目标图像的提取应用,提出了改进Renyi熵的运动目标图像提取方法。首先利用直方图对运动图像的YCbCr颜色概率进行估算,根据图像YCbCr颜色计算Renyi熵,然后设计相应的目标函数来确定阈值,并结合中值处理提高算法的抗噪声性能,最后对颜色的Y、Cb与Cr通道分割得到原始中心点,利用迭代处理更新中心点,并搜索出像素的边界阈值,从而实现运动目标图像的提取。通过仿真,分别从提取时间、计算熵值、以及目标图像识别率多个方面进行比较验证,证明了提出的改进Renyi熵的运动目标图像提取方法能够有效应对运动图像复杂多变的背景,提高了处理速度与提取准确度,具有较好的抗干扰能力。Image extraction of moving object is helpful for image detail analysis under complex changing background through computer vision. Existing image extraction methods usually focus on a certain kind of image, as a result, the application occasion is limited. For this reason, aiming at the application of moving object image extraction, an improved Renyi entropy method for moving object image extraction is proposed. Firstly, the YCbCr color probability of moving image was estimated with histogram, and Renyi entropy was calculated according to the color of YCbCr image. Then the corresponding objective function was designed to determine the threshold, and combining with median processing, the anti-noise performance of the algorithm was improved. Finally, the Y, Cb and CR channels were segmented to get the original center points. Iterative processing updated the center point, and the boundary threshold of the pixel was searched, so that the moving object image can extracted. Through simulation experiments, Comparisons and validations were made in terms of extraction time, calculated entropy and target image recognition rate. It is proved that the proposed improved Renyi entropy method for moving object image extraction can effectively deal with the complex and changeable background of moving image. The processing speed and extraction accuracy are improved, and it has good anti-interference ability.

关 键 词：运动目标 图像处理 中值处理 边界阈值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]