自适应阈值SSDA图像匹配拼接算法的研究

传统的模板匹配算法计算量大，匹配速度慢。序贯相似性检测算法SSDA虽可以减少计算量，但其固定阈值的合适选取存在很大的困难。针对这种情况，提出了一种自适应阈值的序贯相似性检测图像匹配拼接算法。选定阈值初值后，在后续匹配计算中不断自适应更新，遵循超过当前阈值就不再匹配计算的原则，寻尽可能的减少计算量，找最佳匹配点的位置。实例表明，这种算法速度快，匹配准确，图像匹配拼接效果好。

An Algorithm for Images Matching Based on Sequential Similarity Detection with Adaptive Threshold

In the traditional template-matching algorithm,complicated calculations cause a slow matching speed.Though the Sequential Similarity Detection Algorithms(SSDA) may reduce the calculations,it is difficult to select appropriate threshold.An algorithm for images matching is proposed based on SSDA with adaptive threshold in this paper.The initial threshold is set as the sum of color differences of matching regions on the first location.The threshold is updated adaptively when the difference is lower than the current threshold on latter matching location.The calculations are stopped while the difference of a location exceeding the current threshold,the best matching position can be found corresponding to the least difference over whole image.The algorithm can not only get matching position accurately between overlapping images,but also reduce the calculations.The experimental results are satisfactory.