海洋一号C/D卫星(HY-1C/D)海岸带成像仪(Coastal Zone Image, CZI)搭载有2个相机4片CCD，幅宽为950km，在0级原始影像中存在渐晕和片间色差等辐射问题。本文针对海岸带成像仪载荷结构特点，提出了一种针对多片原始CCD影像片间自适应均衡的相对辐射校正方法。首先，采用在轨统计方法对单片CCD影像进行片内辐射校正，明显改善了渐晕问题。然后，通过大视场虚拟重成像方法对多片CCD影像进行高精度几何拼接，在此基础上，统计重叠区域色彩信息。最后，根据相邻片左右重叠区在片间色彩校正前后具有色彩一致性的约束，实现全局多片CCD影像色彩参数解算和片间影像均衡校正。从实验结果来看，经过全视场相对辐射校正处理后，平均行标准差、广义噪声和条纹噪声均优于路径传递法，相对辐射质量优于3%，有效提高了海洋一号C/D卫星海岸带成像仪影像相对辐射质量。

The HY-1 C/D satellite Coastal Zone Imager (CZI) is capable of imaging 950km wide-width through the combination of multiple detector groups, which greatly improves the satellite remote sensing data acquisition capability and is of great significance in global earth observation and ocean water color elements. However, the super-wide imaging system brings severe challenges to the full field of view relative radiometric correction processing: (1) The ground coverage of the push-scan imaging is nearly 1,000 km, which makes the ground calibration field unable to cover all the probes in the imaging field of view; (2) Full field of linear array detectors and the imaging view differences can cause obvious color difference, camera mosaic imaging easily have more serious radiation error accumulation. The CZI is equipped with two cameras and four CCDs for combined imaging. The two CCDs inside the camera adopt collinear full reflection-full transmission optical splicing. Since the mirror will block the overlapping area between the reflection region and the transmission region, energy attenuation will occur at the edge of the image, forming a vignetting. Different focal-plane structure installation and imaging field of view, different CCD, especially the camera response difference is large, resulting in the dynamic light and shade changes between images, there is an obvious color difference. In this paper, an adaptive global color consistency processing method is proposed according to the characteristics of the load structure of the CZI. Based on the single CCD image radiation correction, the color consistency processing can be carried out adaptively by using the global optimization strategy. Global optimization strategy means single CCD image as an unit of all image color correction model processing, to obtain the global meaning of the optimal solution, eliminate the overall color difference between images. Firstly, the on-orbit statistical method is used to correct the radiation of single CCD image, which obviously improves the vignetting problem. Then, the high-precision geometric stitching of multiple CCD images was carried out by using the large-field virtual reimaging method. On this basis, the mean and variance of overlapping area were calculated to describe the color information of each image. Finally, according to the color consistency constraint of adjacent overlapping area before and after color correction, color parameter calculation and image balance correction of global multi CCD images are realized. In the calculation of color parameters, while eliminating the color differences between images, the color differences of the images before and after correction should be ensured as small as possible, so as to increase the number of necessary observation equations, and solve the rank deficit problem caused by the number of overlapping pieces is always less than the number of CCD pieces. According to the experimental results, RA(Root-Mean-Square Deviation of the Mean Line) , GN(Generalized noise method) and Strk(Streaking Metrics) are better than the path transfer method after full field relative radiometric correction, and the relative radiation quality is better than 3%, which effectively improves the image relative radiation quality of HY-1C/D satellite CZI.