一个稳健的用于HDR图像的相机响应函数标定算法

在Mitsunaga和Nayar给出的多项式模型的基础上,增加了图像亮度为0映射到场景亮度也为0的限制条件,并改进了迭代结束条件,所得到的新的标定算法可以使用多张不同曝光度的图像来标定相机响应函数,不同曝光度可以通过改变相机的光圈大小和曝光时间长短来得到.文中所述算法可以自动估计并最后求得精确的曝光度比,省略了原有算法需要输入曝光度比估计值的步骤.算法得到了相机响应函数后,可用多张不同曝光度的图像合成高动态范围的图像.和原有算法相比,此算法的稳健性得到了很大提高,实验表明它对所选的所有输入图像都能收敛,并且得到最优解.

A Robust HDR Image Calibration Algorithm for Camera Response Function

In this paper, a robust calibration algorithm based on a polynomial model is presented. To improve the robustness of the original polynomial model, a new constraint is added so that the pixels with zero brightness are forcedly mapped on to the zero scene radiance. The ending condition of the iterative procedure is also refined in this work, which makes it possible to approach the best solution. The algorithm can calibrate the camera response function from multiple differently exposed images for an arbitrary scene. The different exposures are obtained by changing either the aperture setting or the shutter speed of the camera. The proposed algorithm does not require the exposure ratio of the images; it instead estimates the ratio automatically and can finally get the precise ratio of exposure. With the camera response function obtained, multiple images can be fused into a single high dynamic range radiance image. Compared with the algorithm proposed by Mitsunaga in 1999, the algorithm is more robust and is able to obtain more precise solution. The results show that the algorithm works well for all the images experimented.