基于大数据分析的激光图像模式识别研究  

Research on laser image pattern recognition based on big data analysis

作　　者：胡彦玲[1] 周鹤 HU Yanling;ZHOU He(Nanchang Institute of Technology,Nanchang 330044,China)

机构地区：[1]南昌理工学院,南昌330044

出　　处：《激光杂志》2023年第9期120-124,共5页Laser Journal

基　　金：江西省教育厅科学技术研究项目(No.GJJ212125)。

摘　　要：为降低图像模式识别误差、提高对不同类型图像的识别精度,提出基于大数据分析的激光图像模式识别方法。使用卷积神经网络,对原始激光图像大数据展开训练处理。提取图像特征,并设定阈值对其进行预分割,完成图像预处理。使用正则最小二乘法,完成激光图像模式识别过程,实现图像模式区分。构建实验环节,对此方法的应用效果进行分析。实验结果表明,本方法对于不同类型图像的预设目标图像识别率稳定于97%以上,识别误差率稳定在2%以下,图像识别消耗时间在15 s以内,说明本方法具备较好的应用效果。In order to reduce the error of image pattern recognition and improve the recognition accuracy of different types of images,a laser image pattern recognition method based on big data analysis is proposed.Convolutional neural network is used to carry out training and processing on the original laser image big data.Image features are extracted and the threshold is set for pre-segmentation to complete image preprocessing.Using the regular least square method,the laser image pattern recognition process is completed,and the image pattern differentiation is realized.The application effect of this method is analyzed by constructing experiment.The experimental results show that the recognition rate of the proposed method for different types of images is more than 97%,the recognition error rate is less than 2%,and the consumption time of image recognition is less than 15 s,indicating that the proposed method has good application effect.

关 键 词：模式识别 图像预处理 图像分析技术 激光图像 光谱成像技术 主成分分析 

分 类 号：TN209[电子电信—物理电子学]