摘要
多光源图像合成处理过程中,为了增强图像的轮廓和表面细节,同时使合成后的图像边缘无伪迹、保真度高,提出了一种多光源图像细化与细节增强的协同处理算法。该算法通过梯度域法构建辅助层,采用二次过滤法提取出细节层,提出了新的阴影检测算法用来去除细节层的伪迹,同时使该细节层包含了输入图像的全部信息;使用一幅输入图像构造一个基础层并进行暗区亮度增强处理;细节层和基础层进行复合处理后,重现了暗影区丢失的细节,同时增强了现有的细节。该算法与其他算法进行图像处理对比实验,结果表明,该算法用于实现多光源图像细化和细节增强是可行的、有效的,采用这种方法合成的图像看起来更自然,相对于输入图像,输出图像保持了较高的保真度。该算法具有交互性,用户可以手动或自动调整合成图像的效果。
In order to enhance the image contour and surface details, and make the image edge without artifact after synthe-sis, high fidelity, the authors propose a multi light source image processing algorithm with details enhancement. The algo-rithm, through the gradient domain method to construct assist, secondary filtering method, is used to extract details layer, and a new shadow detection algorithm is used to remove detail layer artifact, at the same time make the detail layer contain all the information in the input image; Using a picture of the input image constructs a base layer and dark space brightness enhancement processing; After detail layer and base layer composite processing, the shadow area missing details are recon- structed. We enhance the existing details at the same time. The results of algorithm and other algorithms for image process-ing experiments show that the algorithm is used to realize the source image and detail enhancement is feasible and effective; using the method of synthetic image looks more natural; relative to the input image, the output image remains high fidelity. The algorithm is interactive; the user can adjust the effect of the composite image manually or automatically.
出处
《重庆邮电大学学报(自然科学版)》
CSCD
北大核心
2014年第2期260-264,共5页
Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)
基金
工业高速缝制装备关键机构智能仿真及可缝性能平台研究(2012C21045)
浙江省科技厅公益项目(2012C20145)~~
关键词
协同图像处理
图像细化
细节增强
阴影检测
collaborative image processing
detail refinement
detail enhancement
shadow detection