针对图像拼接中子图像由于拍摄的角度、相机曝光参数等不同,造成子图像间色调与明暗存在明显差异的问题,提出了一种利用子图像间重叠区域统计信息的色彩校正算法。首先将原始的色彩转换算法利用空间协方差矩阵表述成能直接在RGB空间进行转换的形式,再利用拼接图像间的重叠区域信息对算法进行简化,得到色彩校正算法。实验表明该算法能够有效地应对拼接图像时各子图像间的色调与明暗差异,在图像拼接时可以得到良好的色彩校正效果,同时相比原始的色彩转换算法减少了计算耗时。

A color correction algorithm for image stitching is proposed, which is based on the statistics of the overlap areas between sub-images. The algorithm can handle the problem that there are obvious differences in hue and intensity between sub-images due to the different shooting angles, and exposure parameters in image stitching. Firstly, the original color transfer method is expressed in the form of the transformation in RGB color space with covariance matrices. Then the proposed algorithm is obtained by simplifying that method with the statistics of the overlap areas. The experimental results show that the algorithm can deal with the differences in hue and intensity between sub-images efficiently and make the composite images pleasing looking with color correction. Meanwhile the proposed algorithm is less time-consuming than the original color transfer method.