新的基于边缘特征的图像相关匹配方法

传统的图像相关匹配方法利用求取对应像素灰度差的累加和进行图像之间的相似性度量,或者设定阈值计算具有相近灰度值的像素点的个数进行相似性度量,由于参考图和实测图之间存在着灰度差异、模糊以及局部遮挡等,使得仅依靠图像灰度信息的算法的性能受到影响。结合图像的边界特征匹配技术,提出了一种新的图像相关匹配算法。该算法首先对图像进行边界提取并二值化,然后利用所定义的基于边缘的相似距离(ESD)对处理后的二值图像进行相似性匹配运算。试验结果表明这种算法在多种图像畸变情况下的有效性。

New image correlation-matching algorithm based on edge extraction

The accuracy and validity of the traditional image correlation-matching algorithms are degraded, with the presence of the gray change, image clutter, object variation and occlusion. The reason of degradation is that they use the method of adding the differences of the pixels' gray value according to the corresponding position or calculating the number of the pixels having similar gray value as matching degree. A new image correlation-matching algorithm based on edge extraction is proposed. Firstly the new algorithm processes the image by extracting edge and thresholding, then calculates the correlation of the binary images using the defined ESD distance. Results show that the new algorithm is robust to more image distortions and has great value in target tracking application.