计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (12): 118-124.DOI: 10.3778/j.issn.1002-8331.1903-0145

• 模式识别与人工智能 • 上一篇    下一篇

基于级联神经网络的多任务虹膜快速定位方法

滕童,沈文忠,毛云丰   

  1. 上海电力大学 电子与信息工程学院,计算机科学与技术学院,上海 200090
  • 出版日期:2020-06-15 发布日期:2020-06-09

Multi-task Iris Fast Location Method Based on Cascaded Neural Network

TENG Tong, SHEN Wenzhong, MAO Yunfeng   

  1. College of Electronic and Information Engineering, College of Computer Science and Technology, Shanghai University of Electric Power, Shanghai 200090, China
  • Online:2020-06-15 Published:2020-06-09

摘要:

如何在不受限的实际应用环境下实现虹膜快速定位并具有较高的鲁棒性和准确性是一个非常值得研究的问题,为了快速排除干扰信息准确地定位虹膜区域,提出了基于卷积神经网络的区域和关键点回归多任务虹膜快速定位方法,用矩形框框住虹膜目标区域,在矩形框中用5个关键点定位虹膜区域特征点。实验表明,该方法减少了不受限环境下采集的图像中虹膜内外边界的定位时间和提高了定位的准确性,为下一步虹膜精确分割奠定了良好基础。

关键词: 虹膜定位, 级联, 卷积神经网络, 关键点检测

Abstract:

How to achieve rapid iris location with high robustness and accuracy in an unrestricted practical application environment is a very worthwhile research. In order to eliminate the interference information quickly and locate the iris region accurately, this paper proposes a regional and key points regression multi-task iris rapid localization method based on convolutional neural network. Using rectangle bounding box to encompass iris area and 5 key points to locate iris area feature points. Experiments show that the method reduces the location time of the inner and outer boundaries of the iris in the image acquired in an unrestricted environment and improves the accuracy of location. It lays a good foundation for the precise segmentation of the iris in the next step.

Key words: iris location, cascade, convolutional neural network, key points detection