以某型大面阵彩色电荷耦合器件(CCD)航测相机为研究对象，通过分析其成像过程的各个环节，提出引起相机成像非均匀性的各个因素，并根据分析结果提出光学系统引起的照度差异是导致相机成像非均匀性的主要原因。通过实验室内辐射定标的方法，完成相机暗电流噪声、成像非均匀程度及像元响应度的检测。非均匀性度量结果表明，图像中各彩色像元成像非均匀性均比较严重，尤其是红色像元达到17%，必须予以校正。响应度检测结果显示，相机R、G、B颜色通道各个像元的响应度在工作谱段范围内均呈线性变化趋势，但中心区域像元响应度明显高于边缘区域。通过分析各像元点响应特性的分布规律，提出基于辐射响应特性的非均匀性校正方法，获取各像素点的反演辐射亮度并依据反演辐射亮度的变化，分区域选取进行校正曲线拟合的像素集合。利用最小二乘法获得拟合校正曲线，并对大面阵彩色CCD航测相机的地面成像图像进行了非均匀性校正，其校正效果优于传统的两点校正法。非均匀性度量结果表明，校正后积分球图像的非均匀性降低到4.3%以下，其中红色通道像元非均匀性降低到1.79%，可满足工程实践要求。

All kinds of factors which affect the non-uniformity of a large area array color CCD aerial mapping camera are presented through analyzing the process of camera imaging. According to the analysis results, the differences in illumination of optical system are the main causes for the non-uniformity of camera imaging. The detections and measurements of the dark current noise, imaging non-uniformity and the pixel response are completed through the laboratory radiometric calibration method. The non-uniformity measurement results show that the non-uniformity of each color pixel is serious, especially that of the red pixel reaches 17%, which must be corrected. The detection results of responsivity indicates that the radiometric responsivity of each pixel for R, G and B color channel varies linearly in working spectral range, but the responsivity of pixel in the central area is significantly higher than that of the edge region. By analyzing the distribution of each pixel response characteristics, the non-uniformity correction method based on radiation response characteristic is put forward. Retrieval radiance of each pixel is computed, and according to the variation of radiance, assemblage of pixels which are used to fit the correction curve are selected sub-regionally. The calibration curve is required by using the method of least squares fitting, then the image taken on the ground by area array color CCD camera is corrected, and the correction is better than the traditional two-point calibration method. The measurement results of the non-uniformity show that, after correction, the non-uniformity of integrating sphere image decreases to 4.3% or less, wherein the non-uniformity for red channel pixel is reduced to 1.79%, which can meet the requirements of engineering practice.