Research on Recognition and Location of Stacked Parts Based on Binocular Vision
An efficient identification and positioning method was designed to address the issues of long time and low accuracy in identifying and positioning stacked parts.In terms of recognition,a classifier fu-sion method was proposed.In terms of recognition,a classifier fusion method was proposed.By using this method to train a classifier model with dynamically adjustable weights,the classification and recognition of stacked parts has been successfully achieved.In terms of localization,an improved random Hough transform method is used to extract target edges.Utilize the edge geometric features of stacked parts to reconstruct the spatial pose information of stacked parts.Finally,classification recognition experiments and model evalua-tion were conducted on individual feature classifiers and combined feature classifiers.The experimental re-sults show that the classifier after feature fusion exhibits higher recognition accuracy and more stable accu-racy fluctuations,indicating better classification robustness.
NETL
NSTL
万方数据
