滤光片阵列多光谱图像拼接与灰度调整算法

针对滤光片阵列多光谱图像拼接后单波段图像中地物灰度不一致问题,提出各波段重叠区域灰度均值比的均值统调算法。首先,根据各条带有效区域构建多光谱图像模板,利用平台姿态信息对多光谱图像及图像模板进行投影变换,再利用SURF(Speeded Up Robust Features)算法提取图像匹配点并计算相邻图像间单应性转换矩阵,选择中间序列图像为参考图像,对投影后的图像及其模板进行配准；其次,计算投影后各单波段图像重叠区域灰度平均值比的均值作为图像灰度调整系数,并以参考图像为基准,依次调整配准后图像灰度；最后,利用灰度调整后多光谱图像和配准后图像模板得到各单波段序列图像,通过图像拼接得到灰度一致的大区域单波段图像。理论分析与实验结果表明：该方法不仅可以有效解决滤光片阵列多光谱图像地物灰度不一致的问题,而且能够最大限度保持地物在不同波段下的光谱信息特征。

Filter array multi-spectral image stitching and grayscale adjustment algorithm

To solve the problem of inconsistent grayscale in single-band image after stitching of multi-spectral images of filter array, a new method of the mean tuning algorithm for the mean ratio of grayscale in overlapping regions of each band is proposed. Firstly, multi-spectral image template is constructed based on the effective region of each band, the projection transformation of the multi-spectral image and the image template is carried out using the platform posture information, and the SURF(Speeded Up Robust Features) algorithm is used to extract image matching points and calculate the homography matrix between adjacent images, the intermediate sequence image is selected as the reference image to register the projected image and image template; secondly, the mean value of the ratio of the grayscale averages in overlapping areas of each single-band image after projection is used as the image grayscale adjustment coefficient, and the reference image is used as a reference to adjust the grayscale of the aligned image in turn; finally, each single-band sequence image is obtained by using the grayscale adjusted multi-spectral image and the aligned image template, the large area single-band image with consistent grayscale is obtained by image stitching. Theoretical analysis and experimental results show that this method can effectively solve the problem of filter array multi-spectral image grayscale inconsistency, and then maintain the spectral information characteristics of the features in different wavelength bands to the maximum extent.