摘要
针对高动态范围图像合成过程中相机响应函数标定速度低的问题,提出了一种快速有效的相机响应函数标定的方法.首先根据不同曝光度的图像在成像过程中场景照度保持一致的特性,提出构建一种新的最小平方代价函数,以降低相机响应函数标定方程组的维数;然后在保证标定方程组求解效率的前提下,根据不同曝光度图像的噪音分布,引入高斯加权函数,有效减小多曝光图像在高亮或过暗区域成像过程中引入的噪音,并利用稳健的QR分解算法提高标定方程组的求解速度;最后采用查表的方法,利用高斯加权函数完成场景照度的合成及相应的色调映射,得到能表现整个场景亮区和暗区细节的可视化图像.实验结果表明,与传统方法相比,本文算法在保证相机响应函数标定准确度的同时,具有计算量小、速度快的优点,可实现相机响应函数的快速标定,而且所获得的可视化图像主观效果有所提高,在便携式摄像设备上具有广泛的应用前景.
A fast and effective method is presented to solve the problem of inefficiency in calibration algorithm for camera response function in synthesis of high dynamic range image. Firstly, according to the fact that the corresponding illuminanees on scenes in imaging process among multi-exposure images are accordant, a new least-squares error function is proposed to reduce the dimensions of calibration equation. Gaussian weight function is used in accordance with noise distribution of images in different exposures to eliminate ambient noise in image process in excessive dark and light area with high efficiency to solve equations. Meanwhile, robust QR decomposition is put forward to accelerate the calculation of camera response function. At last, the look-up table method is adopted and Gaussian weight function is used to synthetize the illuminanees on scences and tone mapping; the result image can be well expressed to obtained the detail information of dark and light area in entire scene. The sufficient experimental results on various scenes indicate that this method has smaller computational amount, higher speed than traditional method and has good precision as well. The proposed method can be used to accomplish rapid calibration of camera response function while subjective effect is good in the visualization image obtained by this method. And this method can be applied widely in portable camera equipments.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2013年第6期737-741,共5页
Acta Photonica Sinica
关键词
快速标定
高斯加权
相机响应函数
高动态范围图像
Fast calibration
Gaussian weight
Camera response function
High dynamic range image
