基于改进YOLO的虹膜快速定位检测算法

虹膜定位是虹膜识别系统中不可或缺的环节，针对传统的虹膜定位方法对镜面反射、眨眼等复杂环境下质量差的虹膜图像定位准确率低、计算复杂度高和鲁棒性差等问题，提出了一种基于改进YOLOv3模型的虹膜快速定位方法。针对眼周图像中虹膜内、外圆尺寸变化不大，将YOLOv3网络的多尺度结构改进为双尺度检测；引入了轻量级网络Mobilev3中bneck块来改进特征提取网络，减小模型复杂度；利用K-means++算法对虹膜数据集进行类聚，获得更优的锚点框；模型边框损失函数采用Loss_GIoU改进原均方差(mean squared error, MSE)损失函数；利用虹膜特有几何特征，将模型矩形预测框更改为圆形预测框。在CASIA-IrisV4数据集验证表明，改进模型定位准确率为96.32%,平均精度均值(mean average precision, mAP)为99.37%,检测速度为49.4帧/s,模型参数减少到4.13×10⁶。结果表明改进后的模型较小，并且能够快速精准对虹膜区域定位，具有较高鲁棒性，能够满足虹膜实时定位的场景。

Iris fast location detection algorithm based on improved YOLO

Iris location is an indispensable link in the iris recognition system. Aiming at the problems of traditional iris location methods for poor-quality iris image positioning in complex environments such as specular reflection and blinking, low accuracy, high computational complexity, and poor robustness are proposed. Aiming at the small changes in the inner and outer circles of the iris in the periocular image, the multiscale detection of the YOLOv3 network structure is improved to dual-scale detection; an rapid iris location method based on the improved YOLOv3 model is proposed. Introduced the bneck block in the lightweight network Mobilev3 to improve the feature extraction network and reduce the complexity of the model; using the kmeans++ algorithm to cluster the iris data set to obtain a better anchor box; the model frame loss function uses the improved original mean square error (MSE) loss function; using the unique geometric characteristics of the iris, the model rectangular prediction box is changed to a circular prediction box. Validation on the CASIA-IrisV4 data set shows that the positioning accuracy of the improved model is 96.32%, the mAP is 99.37%, the detection speed is 49.4 frames/s, and the model parameters are reduced to 4.13M. The results show that the improved model is small, can quickly and accurately locate the iris area, has high robustness, and can meet the scene of iris real-time location.