基于MSA特征和模拟退火优化的遥感图像多目标关联算法  

Multi-target association algorithm for remote sensing images based on MSA features and simulated annealing optimization

作　　者：李晖晖[1] 滑立 杨宁[1] 刘坤[2] 

机构地区：[1]西北工业大学自动化学院,西安710072 [2]上海海事大学信息工程学院,上海200135

出　　处：《吉林大学学报（工学版）》2015年第4期1353-1359,共7页Journal of Jilin University:Engineering and Technology Edition

基　　金：航空基金(20131953022);西北工业大学基础研究基金项目(JC20110266);装备研究基金项目(9140A06050113HK****)

摘　　要：由于当前遥感成像技术一般只能获取采样稀疏的遥感图像,无法准确估计目标的状态信息,因此传统的利用状态特征进行关联的方法并不适合遥感图像的目标关联。选取不依赖于时间的目标图像特征作为关联量又无法处理大场景中多个目标关联引起的模糊性。针对上述问题,本文提出了基于多尺度自卷积不变矩特征匹配和模拟退火优化的多目标关联算法。首先提取目标的多尺度自卷积矩(MSA)特征,计算特征间匹配概率,构造整体关联代价矩阵,并设置自适应温度更新函数和双阈值对模拟退火算法进行改进,快速寻求全局最优解。实验结果表明,该算法能够有效地利用遥感图像特征信息,消除关联模糊性,高效解决多目标关联问题。Since the current remote sensing imaging techniques can only get sparse sampling remote images, which are unable to accurately estimate the target state information, so the traditional association method using state features is not suitable for remote sensing image target. If the time independent image features are selected as the associative volume, the ambiguity induced by the association multiple targets in big scene can not be handled. To solve the above problem, a multi-scale autoconvolution feature matching and simulated annealing optimization algorithm is proposed. First, the MSA features are extracted and the matching probability between features is calculated to construct the whole association cost matrix. Then, the adaptive temperature updating function and the double threshold are set to improve the simulated annealing algorithm for quick search of the global optimal solution. Experimental results show that the proposed algorithm can effectively use the feature information of remote sensing image, eliminate association ambiguity and solve the problem of multi-target association efficiently.

关 键 词：摄影测量与遥感技术 目标关联 MSA特征 关联代价矩阵 模拟退火算法 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]