改进的SSDA图像匹配算法

Improved SSDA in Image Matching

在目前的图像匹配中,SSDA实时性好,但对图像灰度的线性变化非常敏感。鉴于此,提出一种基于SSDA(序贯相似度检测算法)的新算法。新算法提出差值矩阵的概念,消除了灰度线性变化的影响。首先将两幅图像同行的相邻像素进行灰度差值计算,获得差值矩阵,再将差值矩阵的元素按照隔点提取的方式进行序贯相似度计算,阈值自适应更新,获得最小阈值的子图像即为匹配图像。实验结果表明,该方法对图像灰度的线性变化有良好的鲁棒性,便于实时性的实现。

The SSDA provides a good real-time performance in image matching, but it is very sensitive to linear transformation of image grey value. In consideration of these facts, a new algorithm for image matching based on SSDA （sequential similarity detection algorithm） is presented. The proposed algorithm can eliminate the gray linear transformation by proposing the concept of difference value matrix. Firstly, the difference values between the adjacent pixels in the same row of two images are calculated to form difference value matrixes. Then the matrix elements extracted alternately are calculated by sequential similarity, to adaptively update threshold, the sub image which obtains minimum threshold is a matching image. Experimental results demonstrate that the al- gorithm has robustness of the linear transformation of image grey values, and facilitates the real-time realiza- tion.