摘要
针对细节丢失和块效应等影响融合图像质量的问题,提出了一种基于Laws特征增强的多聚焦图像融合方法。首先,利用Laws滤波提取多聚焦图像的Laws特征,得到Laws特征图;用改进的均值滤波方法对Laws特征图进行增强处理,得到特征增强图,并对特征增强图采用极值策略提取决策图:最后,将源图像与决策图进行加权融合,得到融合结果。使用该方法对大量图像进行实验处理,结果表明:与其他融合方法相比,基于归一化互信息度量值最大提升32.87%,基于多尺度方案图像融合度量值增加26.57%。该图像融合方法显著提高视觉效果,具有较强的鲁棒性,适用于光场相机的全景深成像。
In response to the problems of detail loss and blocking artifacts that easily affect the image quality during the multi-focus image fusion process,a method based on Laws feature enhancement was proposed.First,we extracted the Laws features of the multi-focus image using the Laws filter to obtain the Laws feature map;we enhanced the Laws feature map using the improved mean filtering algorithm to obtain the featureenhanced map;then we applied the extremum strategy to the feature-enhanced map to obtain the decision map;finally,we performed weighted fusion between the source image and the decision map to obtain the fusion result.Extensive experiments were conducted on a large number of images,and the results showed that,comparing with existing fusion methods,the normalized mutual information metric and the multi-scale scheme image fusion metric had increased by 32.87%and 26.57%,repectively.It can be concluded that the method developed can significantly improve the visual effects with strong robustness,thereby being suitable for all-in-focus imaging of light field cameras.
作者
董晨鹏
郭政义
马怡冉
颜超
章秀华
DONG Chenpeng;GUO Zhengyi;MA Yiran;YAN Chao;ZHANG Xiuhua(School of Mechanical and Electrical Engineering,Wuhan Institute of Technology,Wuhan 430205,China;Hubei Key Laboratory of Optical Information and Pattern Recognition,Wuhan 430205,China;Hubei Engineering Technology Research Center of Video Image and HD Projection,Wuhan 430205,China)
出处
《武汉工程大学学报》
2025年第6期658-663,682,共7页
Journal of Wuhan Institute of Technology
基金
武汉工程大学第十五届研究生教育创新基金(CX2023263)。
关键词
多聚焦图像
图像融合
Laws特征增强
极值策略
multi-focus image fusion
image fusion
Laws feature enhancement
extremum strategy
