摘要
为解决目前工业生产过程中,工件检测技术在复杂环境下特征不清晰且背景复杂情况下识别的不准确等问题,提出了一种多特征融合的HL-S(HOG、LBP特征融合-Softmax分类识别)工件识别算法。首先对获取的图片进行预处理操作,将光照较暗不均匀的图像执行Retinex图像增强,在此操作中使用双边滤波代替高斯滤波;然后进行工件的HOG轮廓特征和LBP纹理特征的提取操作,使用线性融合的方法对处理后的图像进行特征融合;最后,将所有样本数据通过Softmax分类器进行归类判别。实验结果表明,使用该方法进行工件识别,能够提高在光照条件不好情况下的识别率,大大提高了工业生产的效率。
In order to solve the problem of inaccurate identification of workpiece inspection technology in complex environment and complex background, a multi-feature fusion HL-S workpiece recognition algorithm is proposed. The algorithm of this paper first performs pre-processing on the acquired image, and performs Retinex image enhancement on the image with darker and uneven illumination. In this operation, bilateral filtering is used instead of Gaussian filtering. Then, the HOG contour feature of the workpiece and the LBP texture feature extraction operation are performed, and the processed image is subjected to feature fusion using a linear fusion method. Finally, all sample data is classified by the Softmax classifier. The experimental results show that the use of this method for workpiece identification can improve the recognition rate under poor lighting conditions and greatly improve the efficiency of industrial production.
作者
王卉
徐小力
左云波
吴国新
Wang Hui;Xu Xiaoli;Zuo Yunbo;Wu Guoxin(Key Laboratory of Modern Measurement&Control Technology Ministry of Education,Beijing Information Science&Technology University,Beijing 100192,China)
出处
《电子测量与仪器学报》
CSCD
北大核心
2019年第12期94-99,共6页
Journal of Electronic Measurement and Instrumentation
基金
促进高校内涵发展-重点研究培育项目(5211835102)、北京市自然科学基金京津冀基础研究合作专项(J170004)资助。
