航空滤光片阵列多光谱图像条带灰度校正算法  

Grayscale Correction Algorithm for Multispectral Image Stripes of Aerial Filter Array

作　　者：李铜哨 孙文邦[1] 赵汉东 王志磊 白新伟[1] LI Tongshao;SUN Wenbang;ZHAO Handong;WANG Zhilei;BAI Xinwei(School of Aviation Operations and Services,Aviation University of Air Force,Changchun 130022,CHN;Army Logistics Academy,Chongqing 400050,CHN;N0.93525 Troops of PLA,Rikaze 857000,CHN)

机构地区：[1]空军航空大学航空作战勤务学院,长春130022 [2]陆军勤务学院,重庆400050 [3]93525部队,西藏日喀则857000

出　　处：《半导体光电》2022年第6期1173-1178,共6页Semiconductor Optoelectronics

基　　金：全军军事类研究生课题项目(DSSQ910252018010)。

摘　　要：针对航空滤光片阵列多光谱图像中存在的条带灰度差异问题,提出条带灰度校正算法。首先,对多光谱图像构建条带图像模板,基于SIFT(Scale-Invariant Feature Transform)算法提取图像特征点并计算图像间坐标转换矩阵,用于确定条带图像重叠区域;其次,依次计算各条带图像重叠区域像素灰度均值,并以中间图像各条带灰度为基准,依次拟合相邻图像间灰度校正系数;最后,利用投影后的图像模板依次提取序列单波段条带并进行拼接,得到灰度一致的单波段图像。实验结果表明:该方法可以有效解决航空滤光片阵列多光谱图像条带灰度差异,且能够最大限度保持地物光谱信息特征。Aiming at the problem of grayscale difference in multispectral images of aerial filter array, a new method of the strip grayscale correction algorithm is proposed in this paper. Firstly, the strip image template was constructed for multispectral image, and the scale invariant feature transform algorithm was used to extract image feature points and calculate inter-image coordinate transformation matrix, which was used to determine the overlap region of strip image. Secondly, the mean value of the greyscale of pixels in the overlapping area of each strip image was calculated, and the greyscale correction coefficients between adjacent images were fitted with the greyscale of each strip of greyscale intermediate image as a reference. Finally, the projected image templates were used to extract sequential single-band strip in turn and stitch them together to obtain a single-band image with same greyscale. Experimental results show that this method can effectively solve the difference in the greyscale of multispectral image strips of aerial filter array, and can maintain the maximum spectral information characteristicsof the ground feature.

关 键 词：滤光片阵列 多光谱 灰度校正 拟合 单波段 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] TP391[自动化与计算机技术—控制科学与工程]