基于改进Mask R-CNN的光学元件划痕缺陷检测研究

光学元件缺陷会直接影响整个光学系统的性能,在光学元件缺陷检测中,划痕缺陷无疑是检测的难点,划痕缺陷存在着尺寸小,长宽比却比较大,易受杂质影响的问题,本文将深度学习算法应用到光学元件缺陷检测,并根据划痕缺陷的特点,对Mask R-CNN网络模型进行了改进,使算法对划痕缺陷也有了更好的检测效果。首先,将原有的ResNet更换为本文提出的CSPRepResNet,并添加ESE注意力机制,提高了特征提取的能力并减少了计算量;其次,利用K-means算法重新聚类anchor boxes的长宽比例;再次,将目标检测的损失函数由Cross Entropy改为梯度均衡化的Focal Loss,解决了正负样本不平衡问题的同时,更有利于对困难样本的检测,还可以消除离群点的影响。总体来说,检测的mAP@.5由原来的52.1%提高到57.3%,提高了5.2%,且推理速度几乎不变,可见,改进后Mask R-CNN对光学元件划痕缺陷有更好的检测效果。

Research on scratch defect detection of optical elements based on improved Mask R-CNN

Optical element defects will directly affect the performance of the entire optical system. In the detection of optical elementdefects, scratch defects are undoubtedly the difficulty of detection. The scratch defects have the problems of small size, large aspectratio, and easy to be affected by impurities. In this paper, depth learning algorithm is applied to optical element defect detection, andaccording to the characteristics of scratch defects, the Mask R-CNN network model is improved. The algorithm also has a better detectioneffect on scratch defects. First, the original ResNet is replaced by CSPRepResNet proposed in this paper, and ESE attention mechanismis added to improve the ability of feature extraction and reduce the amount of computation. Secondly, K-means algorithm is used torecluster the length width ratio of anchor boxes. Thirdly, the loss function of target detection is changed from Cross Entropy to gradientbalanced Focal Loss, which solves the problem of imbalance between positive and negative samples, is more conducive to the detection ofdifficult samples, and can also eliminate the influence of outliers. In general, the tested mAP@ . 5 The original 52. 1% is increased to57. 3%, an increase of 5. 2%, and the reasoning speed is almost unchanged. It can be seen that the improved Mask R-CNN has a betterdetection effect on optical element scratch defects.