基于语义新型眼部特征的混合参数模型构建  

Building a model that based on semantic features of the new eye mixing parameters

作　　者：刘祥楼[1] 杨龙[1] 孙悦[1] 

机构地区：[1]东北石油大学黑龙江省高校校企共建测试计量技术及仪器仪表工程研发中心,黑龙江大庆163318

出　　处：《光学仪器》2014年第6期533-537,共5页Optical Instruments

基　　金：黑龙江省教育厅科研资助项目(12541055);黑龙江自然科学基金资助项目(F201227);黑龙江省研究生创新科研资助项目(YJSCX2011-121HLJ)

摘　　要：以标准证件照片为研究对象构建了眉毛与眼睛混合一体化参数模型,针对眼睛、眉毛及眉眼共设定14项主特征参数,据此创建28位二进制语义编码。图像处理分为四步:首先,根据眼睛色度信息反映出的眼睛轮廓定位眉毛与虹膜,并针对眼部带状区域实现图像分割;其次,通过改进的Hough变换算法实现对眉毛和虹膜外边缘轮廓检测;然后,用投影方法提取眼部混合特征参数;最后,按特征参量实现语义编码。经1 000张证件照实验证明,图像特征的提取成功率为99%以上,系统对每幅图像自动处理时间小于1.75s。Eye feature extraction is important in the current applications of practical expression recognition, face matching, etc. The traditional eye characterization by semantic description is based on photographs. Firstly, it extracts chrominance eye under eye contour. Secondly, on the basis of the position of the iris and the eyebrows, with edge detection and Hough transform detection, the outer edge of the iris contour is determined. Finally, it extracts mixed characteristics by using the projection method for semantic encoding. This paper analyzes eyebrows and eyes as a whole and a mixed- parameter model is built. The extraction rate is more than 99% in the self-built 1000 passport database. The extraction time is less than 1.75 seconds. It creates a high-resolution semantic description for semantic encoding.

关 键 词：眼部特征 混合参数模型 语义编码 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]