现代电子技术2025，Vol.48Issue(11)：69-76,8.DOI:10.16652/j.issn.1004-373x.2025.11.011
基于深度学习和光学相干层析成像技术的珍珠光泽度分级技术研究
Pearl luster grading based on deep learning and optical coherence tomography
摘要
Abstract
The glossiness of pearls is an important index for pearl evaluating and classification.At present,there are two kinds of methods for classifying pearl glossiness named manual classification and traditional image processing.The former has a slow classification speed and low efficiency,while the latter is prone to environmental interference and has low accuracy.In view of this,a pearl luster grading method based on deep learning and optical coherence tomography(OCT)technology is proposed.After the OCT pearl image acquisition,a convolutional neural network(CNN)model is used to train the collected OCT image data set of pearls,and the trained network is used for prediction.The implementation of the proposed method does not require excessive image preprocessing before the training,improving the efficiency of pearl grading.The comparative experiments on multiple networks has been shown that the ResNet50 model has a high accuracy in pearl glossiness classification,with an average classification accuracy of 96.9%.In comparison with the three classical CNNs,named VggNet16,AlexNet and ResNet18,the proposed method has obvious advantages and achieves rapid grading of pearl glossiness,so it has practical promotion and application value.关键词
光学相干层析成像技术/深度卷积神经网络/珍珠光泽度分类/ResNet50模型/光学特征融合/BatchNormKey words
OCT technology/DCNN/pearl glossiness classification/ResNet50 model/optical characteristic fusion/BatchNorm分类
信息技术与安全科学引用本文复制引用
曹凯,周扬,蔡成岗..基于深度学习和光学相干层析成像技术的珍珠光泽度分级技术研究[J].现代电子技术,2025,48(11):69-76,8.基金项目
浙江省"尖兵""领雁"研发攻关计划(2023C02028) （2023C02028）
