基于Retinex算法的数字航空影像色彩校正研究

针对轻小型无人机数字航空影像的色彩校正问题,笔者采用先匀光,后校正,将原始影像分块,每一块影像都对R、G、B三个通道进行处理,分别在不匀光、Mask匀光、Wallis匀光和Retinex匀光下,基于近距离拍摄的参考影像进行色彩校正,最后合并3个通道及所有分块,消除拼接缝.通过定性分析与定量分析的方法,对比不匀光及3种匀...

Study on color correction based on Retinex algorithm for digital aerial images

Aiming at the color correction problem of digital aerial images with light and small UAVs,the authors adopt the‘dodging first and then correct’method.The original image is firstly divided into blocks,and then each block is processed on the three channels of R,G and B.Based on the reference images taken at close range,the color correction is performed under not dodging,Mask dodging,Wallis dodging,and Retinex dodging,respectively.The three channels and all the blocks are finally merged to eliminate the splicing seam.According to the method of qualitative analysis and quantitative analysis and the comparing between the correction effects of not dodging and three dodging algorithms,it is found that Retinex algorithm is the closest to the reference image.Compared with the original image,the STD and MG deviation is decreased by 44.11%and 78.74%,respectively.By contrast,the results of Mask and Wallis dodging are relatively poor.The color correction after Retinex dodging can best reflect the true colors of the ground objects.The calibration result is better,which is conducive to the modeling and interpretation of aerial images.