灰度相关和区域特征的图像拼接算法

本文提出一种结合灰度特点和区域特征的图像拼接算法。首先采用灰度直方图均衡化的方法降低光照条件不同造成的灰度差异；其次为减少匹配块的计算量，在选取的特征块上计算灰度平均值和每个像素与平均值的绝对差值，然后通过控制阈值来减小搜索范围；最后引入平滑因子消除拼接痕迹，最终实现无缝拼接。实验证明，本文算法合理，可行性强，能够达到良好的拼接效果。

Abstract:

 On the basis of gray correlation and area characters， an algorithm for image mosaic is presented in this paper. Adopting histogram equalization method to reduce grey discrepancy caused by different light conditions firstly. In order to reduce the calculation， the average gray values and the disparity between each pixel and the average values in the selected feature blocks are first computed， and then a threshold is chosen to reduce the search scope. Finally merged overlapped image area data by using smooth factors and then got the whole image without patchlines. The experiments show that the method more effective and more exact for most images.