摘要
虹膜图像质量评估是虹膜识别系统中的重要模块,通过质量评估来摒弃质量较差的虹膜图像,能显著提升虹膜识别系统的性能。虹膜图像质量一般会受到多种因素的影响,单一或少数指标都不能准确客观地进行评估,因此提出了一种新颖的多指标融合的虹膜图像质量评估方法,选取可用虹膜区域、清晰度、虹膜半径、虹膜-瞳孔对比度、虹膜-巩膜对比度、瞳孔扩张性和灰度利用率这7个质量指标,结合GA-BP神经网络进行多指标融合,预测虹膜图像的综合质量评估分数。在CASIA虹膜图像库中进行验证,实验结果表明,该方法可以客观准确地评估虹膜图像的质量,对虹膜识别的性能有很强的可预测性。
The quality evaluation of iris images is an important module in iris recognition system. And the performance of iris recognition system can be greatly improved by abandoning the poor quality of iris images through evaluation. The quality of iris images can be generally affected by a variety of factors. The assessment method using one or two factors cannot evaluate iris images’ quality accurately and objectively, hence a multi-index integration method for evaluating the quality of iris images is proposed. And seven quality factors, including usable iris area, sharpness, iris radius, iris-pupil contrast, iris-sclera contrast, pupil expandability and gray utilization ratio are selected. Multiple factors fusion is combined with GA-BP neural network to predict the comprehensive quality assessment score of iris images. The algorithm is verified in the CASIA data set, and the experiment result shows that the method can select good quality images and has a strong predictability in iris recognition.
作者
晁静静
沈文忠
宋天舒
滕童
CHAO Jing-jing;SHEN Wen-zhong;SONG Tian-shu;TENG Tong(School of Electronic and Information Engineering,Shanghai University of Electric Power,Shanghai 200090,China)
出处
《仪表技术》
2019年第3期24-28,共5页
Instrumentation Technology
基金
国家自然科学基金项目(61772327)
上海市科委地方能力建设项目(15110600700)
