摘要
针对枪械压印字符与其背景色一致和光照条件的影响,通过传统计算机算法进行钢印字符识别时的准确度难以达到理想效果,提出了一种基于YOLOv5的钢印字符识别方法。运用YOLOv5算法提取图像的特征,实现了对钢印字符的识别。实验结果表明,该方法对钢印字符识别的准确率达97.4%,算法平均处理时间为0.016 s,能够满足工程应用的精度和效率要求。此外,利用字符位置信息对模型的输出进行改进,实现直接输出正确的编码信息,在工业生产环境下具有较好的稳定性和实时性,有较强的实际应用价值。
Aiming at the influence of the same color as the background and light condition and there is a problem that the accuracy obtained by traditional computer algorithms for stencil character recognition is difficult to achieve the desired results,a method based on YOLOv5 for recognizing steel embossing characters was proposed.The YOLOv5 algorithm was used to extract the features of the image,thus enabling the recognition of the stamped characters.The experimental results show that the network model achieves an accuracy of 97.4% for the recognition of steel stamp characters,and the average processing time of the algorithm is 0.016 s,which meets the accuracy and efficiency requirements for engineering applications.Besides,the output of the model was improved by using character position information to achieve direct output of the correct coding number.With good stability and real-time performance in an industrial production environment,this has some application implications.
作者
宫鹏涵
GONG Penghan(Shijiazhuang Campus of Army Engineering University,Shijiazhuang 050000,China)
出处
《兵器装备工程学报》
CAS
CSCD
北大核心
2022年第8期101-105,124,共6页
Journal of Ordnance Equipment Engineering
