彩色宽幅面阵航空相机的非均匀性校正方法

分析某型Bayer阵列彩色宽幅面阵航空相机的成像过程,结合Bayer阵列成像原理,提出相机成像均匀性的各个影响因素。采用实验室内辐射标定的方法获得相机暗电流噪声及像元响应度的测量数据,响应度测量结果表明,在工作谱段范围内,当输出值未达到饱和时,相机各色彩通道像元的响应度曲线接近线性变化,但像面边缘区域的像元响应度较中心区域明显降低。以响应度最高的红色通道为基准,获得等能量输入时各色彩通道的归一化色彩平衡系数,结合各像元响应特性,提出基于辐射亮度反演和等能量色彩平衡系数的非均匀性校正方法,并对相机航空成像图像进行了非均匀性校正,均值、方差及梯度等统计结果表明,校正后图像各个区域各个色彩通道的灰度均值差异明显减小,尤其是红色通道,其左下区域标准方差比由0.201 5降低到0.076 1,说明图像整体亮度更均衡,色调更均匀,校正后图像各个色彩通道的平均梯度较原始图像有所提高,说明校正后图像细节信息更加丰富。

Non-uniformity correction method for Bayer array color wide format aerial camera

Based on Bayer array wide format color aerial camera , imaging process of the camera is analyzed , and combined with the imaging principle of Bayer array , all kinds of factors which affect the non-uniformity of the camera are presented.The detection and measurement of the dark current noise and the pixel response are completed through the radiometric calibration in laboratory .The measurement results of responsivity indicate that in unsaturated state , the radiometric response characteristics of each pixel for each color channel vary linearly in working spectral range , but the responsivity of pixel in the edge region is significantly lower than the central area .The R channel with the maximum absolute spectral sensitivity is chosen as the reference channel , and the normalized color balance coefficient is acquired for each color channel.Combined with the distribution of each pixel response characteristics , non-uni-formity correction method is proposed based on the radiation brightness inversion and equal-energy color balance coef-ficient, and the aerial image is corrected.The statistical results of mean, variance and gradient show that the differ-ence of mean gray of each area and each color channel for corrected image are significantly reduced , especial for the red channel.The standard deviation ratio on left lower area is decreased from 0.201 5 to 0.076 1, the overall image brightness is more balanced , and color is more uniform .Compared to the original image , the average gradient of each color channel is improved , and it shows that the detail information of corrected image is more abundant .