光学仪器, 2014, 36 (6): 533, 网络出版: 2015-01-13
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
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1.75 s。
Abstract
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 1 000 passport database. The extraction time is less than 1.75 seconds. It creates a high-resolution semantic description for semantic encoding.
刘祥楼, 杨龙, 孙悦. 基于语义新型眼部特征的混合参数模型构建[J]. 光学仪器, 2014, 36(6): 533. LIU Xianglou, YANG Long, SUN Yue. Building a model that based on semantic features of the new eye mixing parameters[J]. Optical Instruments, 2014, 36(6): 533.