摘要
针对普通相机拍摄图像中存在区域曝光问题,提出一种改进的基于离散余弦变换的多曝光图像融合算法。分别对同一场景的多幅不同曝光图像分块,并对每个小块进行离散余弦变换。对提取相应位置的低频系数平均加权,获取图像近似部分的融合分量;对高频系数基于区域标准差确定权重系数,根据权重系数的大小获取细节部分的融合分量。最后,通过反离散余弦变换获得较好质量的多曝光融合图像。仿真实验表明,该算法在提取低频系数为25%,高频系数为75%时,展现的图像细节信息更加丰富。
There are overexposure or under-exposure problems in images on ordinary camera,and these result in poor image quality.An improved multi-exposure image fusion algorithm is proposed based on DCT transform to relieve the above problem.Firstly,multi-exposure images of the same scene are respectively divided into blocks and DCT transform is applied to each of block.Secondly,each block is extracting the low-frequency coefficient and high-frequency coefficient.For low-frequency coefficients,they are average weighted and used as the approximation of the fusion image.For high-frequency coefficients,the local standard deviation of every pixel is calculated to achieve the weights of every high-frequency coefficients,and then weighted average the coefficients.Lastly,inverse DCT transform is applied to obtain the fusion one of the multi-exposure images.Simulation results show that the algorithm can provide richer details of the image while extracting the low-frequency coefficient of 25%and the high-frequency coefficient of 75%.
出处
《西安邮电大学学报》
2016年第6期40-43,共4页
Journal of Xi’an University of Posts and Telecommunications
基金
陕西省国际科技合作计划资助项目(2015KW-014)
关键词
图像融合
多曝光图像
DCT变换
局部标准差
加权平均
image fusion
multi-exposure image
DCT transform
local standard deviation
weighted average