克服动态问题影响的相机响应函数标定

利用图像之间的亮度映射函数而不是图像本身各像素点信息进行相机响应函数的标定,避免了成像系统的晃动或者拍摄场景的动态问题带来的图像配准之间的误差给后续标定算法带来的影响。采用直方图规则化的方法分析不同曝光量图像之间的各颜色通道不同亮度级的统计特征,获得了不同图像对之间的亮度映射函数。然后,利用图像对之间的亮度映射函数结合各帧图像的直方图建立求解相机响应函数的超定方程模型。最后,利用最小二乘法解算模型获得相机响应函数。验证实验表明,本文相机响应函数标定算法具有克服相机的动态问题的能力;不同输入帧数图像的分组实验显示,最小输入4帧图像能够获得较为理想的响应函数标定结果。

On calibration of camera response function free from the impact of dynamic problem

Calibration of camera response function was carried out by means of brightness mapping between images rather than pixel information of the images themselves, which could help avoid the impact of image registration errors caused by image system shaking or dynamic problem in the shooting site on the follow-up calibration algorithm. The method of histogram regularization was adopted to analyze the statistical characteristics of different color channels with different light levels between images of different exposures, and the brightness mapping function between different image pairs were obtained. Then, the over-determined equation model was worked out by combining the brightness mapping function between different image pairs with the histogram of each frame image. For the last step, the camera response function was obtained by means of the least square method. It shows that the calibration algorithm for camera response function can help overcome the impact of dynamic problem, and the grouping experiment on images with different input frames indicates that the fewest images with 4 input frames can achieve the optimal camera response function calibration value.