机器视觉在叶片编码自动识别中的应用研究  

Research on the application of machine vision in blade coding automatic recognition

作　　者：张超 徐微雨 李灿伟 毕超[1] 胡成海[1] ZHANG Chao;XU Weiyu;LI Canwei;BI Chao;HU Chenghai(Beijing Precision Engineering Institute for Aircraft Industry,Beijing 100076,China)

机构地区：[1]北京航空精密机械研究所,北京100076

出　　处：《兵器装备工程学报》2023年第S02期189-193,共5页Journal of Ordnance Equipment Engineering

基　　金：国家自然科学基金项目(52175491)。

摘　　要：提出了一种基于机器视觉和深度学习的方法,可快速识别叶片底部编码。利用机器视觉和图像处理技术提取叶片编码区域,构建卷积递归神经网络字符特征,构建582个样本字符和349个通用字符的训练集预训练字符识别模型,使用包含1752个字符的匹配模板作为实验的初始条件进行模板匹配。经过人工干预后,识别率达到98.92%。研究结果表明:图像处理技术和深度学习、模板匹配算法相结合在飞机发动机叶片编码识别方面有巨大的实用价值,为类似应用场景提供可靠解决方案。Aero-engine blades need to be identified by coding in production.In order to solve the problem that manual identification of coding is not only a waste of manpower,but also has a high error rate,this paper proposes a method based on machine vision and deep learning,which can quickly identify the bottom code of the blade.Using machine vision and image processing technology,the leaf coding region is extracted,and a convolutional recurrent neural network is constructed for character features.A training set of 582 sample characters and 349 general characters is constructed to pre-train the character recognition model.After manual intervention,the recognition rate reaches 98.92%.The results show that the combination of image processing technology,deep learning and template matching algorithm has great practical value in the coding and recognition of aircraft engine blade,and provides reliable solutions for similar application scenarios.

关 键 词：机器视觉 图像处理 深度学习 模板匹配 航空发动机 

分 类 号：TN919[电子电信—通信与信息系统]