激光与光电子学进展, 2020, 57 (21): 211407, 网络出版: 2020-11-09  

基于机器视觉的激光表面改性质量快速检测 下载: 787次

Rapid Detection of Laser Surface Modification Quality Based on Machine Vision
田崇鑫 1,2李少霞 1,2虞钢 1,2何秀丽 1,2王旭 1,2
作者单位
1 中国科学院力学研究所, 北京 100190
2 中国科学院大学工程科学学院, 北京 100049
摘要
针对铜铬合金激光表面改性质量快速无损检测的需求,提出了一种基于机器视觉的检测算法。首先采集试样表面形貌图像,然后使用自适应二值化方法从背景图像中分割视觉显著区域,再基于几何矩提取具有空间变换不变性的连通域形状特征,最后依据激光能量输入定义4种基本改性状态并训练支持向量机,以检测改性质量。使用MATLAB语言实现上述算法,结果表明:本文算法在特征提取及模型训练阶段的耗时约为45 s,检测速度为5×10 6pixel/s,检测准确率为97.0%。依据检测结果可进行相应的工艺参数优化。所提算法对光照等检测环境不敏感,可以实现激光表面改性质量的快速无损检测,且对工艺参数优化具有一定意义。
Abstract
In this study, a method based on machine vision is proposed for the rapid nondestructive detection of laser surface modification in copper-chromium alloy. Surface morphology images of the specimen are collected, and the visual salient regions are segmented from the background by applying the adaptive thresholding method are extracted. Additionally, based on geometric moments, the characteristics of the connected domain with spatial transformation invariance. According to the laser energy input, four basic modification states are defined, and a support vector machine is trained to determine the modification quality. Writing scripts in MATLAB language, the results show that it takes about 45 s for feature extraction and model training. Moreover, the recognition speed is about 5×10 6 pixel/s, and the recognition accuracy is about 97.0%. Based on the detection results, the corresponding process parameters can be optimized. Furthermore, the method is not sensitive to light and other detection environment factors, thereby achieving the requirement of rapid and nondestructive detection of laser surface modification quality, which has a certain significance for the optimization of process parameters.

田崇鑫, 李少霞, 虞钢, 何秀丽, 王旭. 基于机器视觉的激光表面改性质量快速检测[J]. 激光与光电子学进展, 2020, 57(21): 211407. Tian Chongxin, Li Shaoxia, Yu Gang, He Xiuli, Wang Xu. Rapid Detection of Laser Surface Modification Quality Based on Machine Vision[J]. Laser & Optoelectronics Progress, 2020, 57(21): 211407.

引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!