基于边缘几何特征和频域相关技术的图像匹配方法

Image matching based on geometric feature of edges and the correlation in frequency domain

提出了一种基于图像特征提取和变换域相关的图像匹配方法。借助提取图像的大边缘信息,以其几何特征构造边缘方向角曲线作为图像匹配的依据。通过对差分的计算,并比较两幅图像边缘方向曲线的相对链码,得到2幅图像之间可能的旋转角度。对旋转矫正后的图像进行2维傅里叶变换,计算频域内2图的相位相关矩阵,并从中找到相关系数最大的位置,这个位置就代表了2图的坐标位移量,通过平移达到图像间的精确匹配。实验表明:该方法匹配准确,速度快,减少了人为因素的干扰,对于图像间存在较小灰度差和噪声情况均有很好的匹配效果。

The new method of image matching based on feature extraction and transform domain correlation is proposed in this paper. By extracting long edge information from image and analyzing its geometric feature, the edge direction-angle curve can be constucted as an important parameter for image matching. By calculating difference and comparing with relative chain codes of two images＇ edge direction curves, the possible rotation angle between the two images will be obtained. After rotating images and transforming them with 2D Fourier, the phase correlation will be gained in frequency domain and the maximum of the correlation coefficient will be found in that position which indicates the displacement of two images. So the two images can be matched accurately. Experimental results prove that this matching method is accurate, fast and automatic, it decreases the disturbance caused by artificial factors and could achieve a satisfactory matching between the two images with low grey difference and noise.