基于矢量加权线检测算子的虹膜环状线条检测

Iris annular line detection based on vector weighted detection operator

由于虹膜环状线条与背景相比信号强度比较弱,且背景图像纹理异常丰富,环状线条灰度不完全连续,在灰度图像上直接用边缘检测算子进行虹膜环状线条检测时,会丢失很多有效信息,误检率及漏检率较高,为了解决这个问题,在对虹膜环状线条特征进行分析的基础上,设计了一种适用于虹膜环状线条检测的矢量加权线检测算子。首先由于环状线条在虹膜上的位置比较固定,选取虹膜环状区域左右两部分区域作为感兴趣区域;其次,设计矢量加权线检测算子,依据优势信号出现在不同通道的随机性自适应的进行加权,由此将矢量图像转变为边缘信息最为突出的单通道图像,再根据环状线条的线特征分布特点设计线检测矩阵,对预处理后的感兴趣(ROI)区域进行检测;最后在二值化图像上根据环状线条的形状设计区域形状因子排除非环状线条,实现环状线条的检测。该方法在图库中人工标定的1 921条环状线条的检测正确检出率达到91.78%。实验结果表明,本方法与经典的边缘检测算子相比,更适用于虹膜环状线条的检测。

Iris annular line detection in the gray image directly with the edge detection operator will lost a lot of useful information, and false rate and missing rate is high, due to the signal intensity of iris annular line is weak compared with the background and the background image texture is very rich and the gray value of the annular line is not completely continuous. Thus, an iris annular line detection method is presented in this paper on the basis of analyzing the features of the iris annular lines. Firstly, the two part of the annular areas are selected as the region of interest due to the annular line position is fixed in the iris image; Secondly, the vector weighted line detection operator is designed to transform the random nature of the dominant signal into the adaptive weighted value, and the vector image is transformed into the single channel image with the most prominent edge information. Afterwards, the line detection matrix is designed to detect ROI according to the line characteristic distribution of annular line. Finally, non-annular lines are eliminated by shape factors of the binary images and the annular lines are detected. The method is used to detect 1921 lines that were artificially marked in our gallery, and the detection accuracy is up to 91.78%. The experimental results show the proposed method is suitable for detecting the iris annular lines from the visible iris images compared classic edge operators.