Research On Vehicle Hub Type Recognition System Based On SURF-LBP Feature Integration

With the successive proposal of Industry 4.0 and Made in China 2025,intelligent manufacturing has become the development direction of the industrial manufacturing field.Machine vision technology,pattern recognition technology,and artificial intelligence technology represented by machine learning have developed rapidly,and it has been more and more applied to actual production and manufacturing,and has demonstrated excellent performance beyond conventional methods in recent years.In order to improve the level of automated vehicle hub production and replace the way of manually recognizing the type of the hub,this topic applies machine vision technology and pattern recognition technology to identify the type of vehicle hub,and has conducted an in-depth study on the method of vehicle hub identification.The main work and conclusions of the thesis are as follows:(1)In order to obtain the region of interest of the vehicle hub image,so as to avoid the influence of some background noise and reduce the calculation amount of subsequent algorithm processing.Three aspects of image filtering,image threshold segmentation,and binary morphological operation were studied.Several commonly used filtering algorithms,threshold segmentation algorithms,and binary morphology algorithms were theoretically discussed and experimentally analyzed,and the corresponding processing algorithms and operating parameters were determined by comparison.Finally,using the methods of contour search and fitting,and rectangular boundary approximation to obtain the region of interest of the hub image.(2)By analyzing the characteristics of each type of hub,it is clear that the characteristics of the hub need to be comprehensively expressed from the perspective of shape and texture.In order to take into account the accuracy,robustness and the rate of recognition,the SURF feature of the hub image is extracted to reflect the shape information of the hub,and the LBP feature of the hub image is extracted to reflect the texture information of the hub;For the SURF feature and the LBP feature,the paper proposes a method that a feature bag based on SURF features using bag of feature algorithm is constructed directly,and a feature bag based on LBP features isconstructed using image cell division and bag of feature algorithm,and then integrate the two in series.Finally,the two features are integrated into an image representation by the proposed method,which lays the foundation for the next training of classification model.(3)The SVM classification algorithm and BP neural network classification algorithm are studied.Aiming at the problem that the traditional SVM classification algorithm cannot perform multi-class classification,an error correction output coding framework is used to construct an ECOC multi-class SVM classification model suitable for multi-class classification.In order to verify the effectiveness of the SURF-LBP integrated features,four types of hubs Q19,Q12,R97,O66 in the hub library were selected,and corresponding sample image sets were created.The SURF features and integrated features of the sample image sets were used to train their respective ECOC multi-class SVM classification model and BP neural network classification model.Through experimental comparison and analysis,the excellent performance of integrated features is proved,and it is determined that the BP neural network classification model based on integrated features has higher classification accuracy.(4)Through the construction of software and hardware systems,the design and implementation of the vehicle hub type recognition system was completed,and the effectiveness and advancedness of the system were verified through the operation.In summary,on the basis of machine vision technology,this topic has conducted an in-depth discussion on the method of identifying vehicle hub types using pattern recognition.A recognition algorithm with high recognition accuracy,strong robustness,and less time-consuming is proposed.A set of vehicle hub type recognition system is set up based on the recognition algorithm,which realizes the automatic recognition of the hub type,and improves the level of automation and intelligence of the entire hub production line.