计算机辅助虹膜诊断中特征提取方法研究  

Study on feature extraction in computer-aided iridology

作　　者：辛国栋[1] 王巍[2] 

机构地区：[1]哈尔滨工业大学软件学院,山东威海264209 [2]哈尔滨工业大学计算机科学与技术学院,山东威海264209

出　　处：《计算机工程与设计》2006年第18期3322-3323,3376,共3页Computer Engineering and Design

基　　金：国家自然科学基金项目(90209020)。

摘　　要：针对计算机辅助虹膜诊断中虹膜分区以及局部特征提取问题,提出了虹膜区域划分方法及具体分区中特征提取的方法。利用虹膜图像中瞳孔的灰度特征,实现虹膜的快速定位。在虹膜定位的基础上,利用虹膜图谱对左右虹膜图像分别进行特征区域划分,并且对划分后的虹膜图像进行规一化处理。在规一化后的虹膜图像中,针对肠区的病理特点,利用改进的Hough变换,提取了虹膜图像中肠道分区的局部病理特征。实验表明,虹膜图像区域划分方法的可行性。To solve the problems of iris image partition and feature extraction on local iris image in the computer-aided iridology system, the iridology diagram mapping method and given area pathology feather extraction method are presented. Iris location methods relying on eye gray level is applied in the process. After finding the pupil boundary and combined with the circle template, the outer boundary of the iris is found. On the result of iris location, iris image is divided into many small local areas corresponding to the iridology diagram in order to make the iris and local iris image area comparable to the same iris of different period, In order to illuminate the process of local area feather extraction in the local area, feature in the intestine area is extracted in the located iris image using improved Hough transform.

关 键 词：图像处理 虹膜诊断 虹膜定位 HOUGH变换 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]