海面舰船目标检测优化算法仿真

The Simulations of an Optimal Algorithm for Sea Surface Ship Targets Detection

对于海面舰船目标图像受海空背景影响,容易在检测过程中出现外点和误检的问题,评测各类Harris角点检测改进方法,并结合极差方法,构建了一种协同多角度最优改进策略的特征点检测稳健算法。利用Harirs检测算法计算复杂度低的优势,分别从滤波器改进、自适应阈值、临近点剔除以及亚像素定位四个角度出发,筛选出每个角度下典型的优化方法,并在相同的试验条件下,分析对比各种优化方法的性能,从而得出最优的改进策略。基于极差的方法,可以对海空背景进行有效的剔除,最终给出了多角度下的协同优化算法。仿真结果表明,在同等试验条件下,上述算法的综合性能优于以往任一单角度的改进算法,可有效滤除海空背景杂波,检测出目标图像。

In the paper, we improve Harris corner detection algorithm with global optimization. After testing sev- eral modified Harris algorithms, a feature point corner detection algorithm of collaborating multi-angle optimal im- provement strategies was constructed combining with the range method. From the views of filter modification, adaptive threshold, neighboring point excluding and sub-pixel positioning, we selected the typical optimal methods, then com- pared the results of these methods under the same test conditions and obtained the optimal modified strategies. To in- tegrate these strategies can derive the multi-angle collaborative optimization algorithm. Numerical resuhs show that the proposed algorithm is better than past modified ones. Filtering the sea and air clutter, images targets can be de- tected.