基于对数极坐标变换的快速相关匹配算法

成像制导过程中,需要在导引头获得的图像中搜索目标所在位置。基于对数极坐标变换的相关匹配算法能够有效处理在尺度、旋转等情况下目标的匹配问题,但是计算量大、匹配效率低,因此提出一种基于对数极坐标变换的多分辨率快速相关匹配算法。首先,通过定义主梯度方向,并将该方向作为对数极坐标变换的角度变换基准,有效避免了模板与目标特征矩阵基准不同的问题;其次,在多分辨率金字塔搜索过程中,采用图像方差作为判断依据,跳过同质区域和杂乱区域等非目标区域,加快搜索效率。仿真结果表明,该算法相比传统多分辨率相关匹配算法能够保证匹配精度并提高算法对于非线性光照变化的鲁棒性,匹配效率显著提高,具有实时处理的潜力。

A fast correlation matching algorithm based on log-polar transform

In the imaging guidance,it is necessary to locate the target in the image acquired by the seeker.The correlation matching algorithm based on log-polar transform can effectively deal with the matching problem when target has changed in scale or rotated,while the computation is too expensive and the efficiency is low.Therefore,a fast multi-resolution correlation matching algorithm based on log-polar transform is proposed in this paper.Firstly,the principal gradient direction is defined and chosen as the starting direction of the angle axis of the log-polar transform,so as to avoid the problem of different starting directions between the feature matrices of template and target.Secondly,in the process of multi-resolution search,the variance of sub-image is used as the judgment to determine whether this sub-image is to search or skip,which could effectively skip non-target areas such as homogeneous regions and cluttered regions,and speed up the matching efficiency.The experimental results show that the proposed algorithm can not only guarantee the matching accuracy but also improve the robustness of the non-linear illumination changing.At the same time,the matching efficiency is significantly improved,and the proposed method has the potential of real-time processing.