基于超像素分割的无线内窥镜出血图像检测  

Detection of bleeding image in wireless capsule endoscope based on superpixel segmentation

作　　者：付延安[1] 孟庆虎[1,2] 张伟[1] 

机构地区：[1]山东大学控制科学与工程学院,济南250061 [2]香港中文大学电子工程系,香港999077

出　　处：《吉林大学学报（工学版）》2013年第2期510-514,共5页Journal of Jilin University:Engineering and Technology Edition

基　　金：国家自然科学基金项目(61203253);山东大学自主创新基金项目(2012ZD016)

摘　　要：将图像形态学处理和计算机辅助诊断方法应用于无线内窥镜出血图像检测。在对无线内窥镜图像进行形态学处理和超像素分割的基础上,提出红色纯度特征,应用最小二乘支持向量机分类器对分割后的图像进行分类。试验结果表明,该方法能够有效地检测无线内窥镜图像中的出血区域,敏感性、特异性和准确率分别达到了99%、94%和95%,可以作为临床医生的辅助诊断工具。The morphological processing and the computer-aided diagnosis were used to detect the bleeding images in wireless capsule endoscope（WCE）. Based on the morphological processing and the superpixel segmentation of the WCE images, the red ratio color feature was proposed, and the segmented images were classified by the least squares support vector machine. The experiment results showed that the proposed method can effectively detect the bleeding regions in WCE images, its sensitivity, specificity, and accuracy can reach 99%, 94%, and 95% respectively. This method can help the clinician examine the gastrointestinal diseases.

关 键 词：信息处理技术 无线内窥镜 出血检测 图像分割 超像素 最小二乘向量机 

分 类 号：TN911.73[电子电信—通信与信息系统]