基于互相关信息的图像目标检测方法研究  

Studies on Image Target Detection Method Based on Mutual Information

作　　者：张永宏[1] 李建楠[1] 

机构地区：[1]南京信息工程大学信息与控制学院,江苏南京210044

出　　处：《计算机仿真》2013年第6期397-400,共4页Computer Simulation

摘　　要：针对背景变化下运动目标的检测问题,采用一种结合互相关信息的目标检测算法,利用混合优化,Powell和粒子群(Pso)交替优化两幅图像间的互相关信息,直到互信息值最大,得到图像间的最佳平移旋转参数,将图像进行空间配准。最后采用三帧差分相乘法检测出目标并增强目标区域,实现运动目标的准确检测。通过对变化背景图像的运动目标检测仿真,结果证明了检测算法的准确性和可行性。Against the moving object detection in the background of changes, this paper proposed a target detection algorithm combined with the mutual information. The algorithm uses a hybrid optimization of the Powell and PSO to optimize the mutual information between the two image information alternately till the mutual information of the maximum value, and gets the best translation rotation parameters in the image then match the images. Finally, it u- ses inter - frame difference multiplication to detect the target, enhance the target area, and detect moving targets suc- cessfully. The experimental results demonstrate the accuracy and feasibility of this algorithm.

关 键 词：互相关信息 空间变换 混合优化策略 目标检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]