摘要
针对传统的图像识别算法识别多品种工件,存在运行时间长、识别率低等问题,提出基于改进ORB-SVM的工件图像识别方法。在传统ORB算法的特征点检测基础上,采用SIFT算法获得具有旋转尺度不变性的图像特征描述,并利用词袋模型将图像特征转化为特征直方图,进而构建支持向量机(SVM)的分类模型,实现对工件的识别分类。试验结果表明:改进的ORB-SVM在应对旋转变换、光照变换、尺度变换时更具鲁棒性,且工件识别准确率高达98.89%,单个工件的识别时间低于0.43 s,具有良好的高效性和实用性。研究为多领域的工件识别提供参考。
In order to address the problems of long running time and low recognition rate in traditional image recognition algorithms for identifying multiple types of workpieces,a workpiece image recognition method based on the improved ORB-SVM was proposed.The SIFT algorithm was employed based on the feature point detection of the traditional ORB algorithm to obtain image feature descriptions with rotation and scale invariance.The image features were then transformed into feature histograms using the bag-of-words model.Subsequently,a classification model based on support vector machines(SVM)was constructed to achieve workpiece recognition and classification.Experimental results show that the improved ORB-SVM exhibits robustness in handling rotation,illumination and scale transformations.Moreover,the workpiece recognition accuracy reaches 98.89%,with the recognition time of less than 0.43 seconds per individual workpiece,indicating that the method has high efficiency and practicality.This research provides a reference solution for workpiece recognition in various fields.
作者
仝保国
刘凌云
TONG Baoguo;LIU Lingyun(School of Electrical&Information Engineering,Hubei University of Automotive Technology,Shiyan442002,China)
出处
《包装与食品机械》
CAS
北大核心
2024年第1期60-66,共7页
Packaging and Food Machinery
基金
国家自然科学基金项目(51575211)
湖北省自然科学基金项目(2016CFB401)
湖北省教育厅科学技术研究项目(Q20201801)
湖北汽车工业学院博士科研启动基金项目(BK202004)。
关键词
工件识别
特征检测
ORB算法
词袋模型
支持向量机
workpiece identification
feature detection
ORB algorithm
word bag model
support vector machines
