机电工程技术2025，Vol.54Issue(9)：115-118,178,5.DOI:10.3969/j.issn.1009-9492.2025.09.022
基于神经网络的机器人工件识别与分类研究
Research on Robot Workpiece Recognition and Classification Based on Neural Networks
摘要
Abstract
In response to the complex challenges encountered in the assembly and production of robot joint workpieces,including a wide variety of workpieces,large production scale,and low efficiency in manual sorting and assembly,A cutting-edge industrial robot workpiece recognition and classification algorithm based on neural networks is proposed.The core process of this algorithm starts with high-definition workpiece images captured by industrial cameras.Then,through advanced computer vision technology,these images are quickly read and processed through a series of fine preprocessing steps,including image grayscale to simplify information complexity,mean filtering to eliminate noise interference,and adaptive threshold segmentation to accurately extract key features,ultimately generating clear and accurate binary images.Finally,the algorithm accurately extracts the contour of the workpiece in the binary image,and use Hu's moment invariants to accurately compare the contour of the workpiece to the pre-set template contour for workpiece recognition and classification.The experimental results verify the excellent performance of the algorithm in workpiece recognition and classification tasks,with a workpiece recognition accuracy of 100%.Not only does it have high recognition accuracy,but it also has excellent robustness and adaptability,providing a valuable reference scheme for similar recognition tasks.关键词
工件分类/机器视觉/神经网络Key words
workpiece classification/machine vision/neural network分类
计算机与自动化引用本文复制引用
刘杰,张松亚..基于神经网络的机器人工件识别与分类研究[J].机电工程技术,2025,54(9):115-118,178,5.基金项目
湖南省社会科学成果评审委员会课题(XSP24YBC484) （XSP24YBC484）
