中文摘要：

在铜带表面缺陷检测系统中,针对仅硬件改善缺陷图像精细特征信息,受制造水平、成本等因素制约以及传统超分辨复原方法实时性不强等问题,提出一种基于粗糙集（RS）与纹理特征预分类的快速超分辨率（SR）图像复原方法。本文方法利用RS属性约简原理,选择并优化对弱纹理缺陷目标描述性较好统计特征参数,并在匹配搜索时根据纹理特征对样本库进行预搜索分类,然后在分类得到的纹理内容相近的样本子集中对输入的低分辨率（LR）样本块精确匹配搜索。理论和实验结果表明：本文方法应用于铜带缺陷在线检测系统中,可使缺陷区域的高频信息增强、边缘和细节更加清晰,且算法实时性较好,在兼顾图像复原质量和运行效率上具有优越性和可行性;并可用于其它金属表面的图像复原。

英文摘要：

In the conventional copper strip surface defect detection system,there exist problems that hardware improvement for more detailed information in captured images is confined by factors such as manufacturing level and costs,and the conventional super-resolution（SR）restoration method has poor realtime performance.In order to resolve these problems,we propose a fast super-resolution restoration method based on rough set and pre-classification of texture features in this paper.First of all,by using the theory of attribute reduction approach of rough set,we select and optimize those statistical characteristic parameters which offer better descriptions on targets with tiny texture defects.And at the same time,according to the texture features,we make pre-search and pre-classification on sample sets when doing matching search.Finally,in the pre-classified sample subsets with similar texture features,we find exact matches for the inputs of low-resolution images.Both theoretical analysis and experimental results demonstrate that our method is able to enhance high frequency information in defect areas,sharpen edges and details,and perform better in real-time systems,and also,it has superiority and feasibility in balancing quality with efficiency in image restoration when applying it in the online system for copper stripe surface defect detection.