Research On Vehicle Parts Defect Detection Based On SVM And HOG

At present,automobiles occupy a very important position in people's lives.The appearance of automobiles brings enormous convenience to people's lives.Subsequent safety problems have also emerged in people's lives.In order to ensure the safety of the car,it is essential to inspect the assembly of each link in the production process.In the early days,some workers were responsible for the inspection of vehicle parts assembly,but at present,simulation inspection system is widely used in factories to complete part of the manual work.However,due to the inherent problems of the inspection mechanism,the inspection accuracy is low,which still needs to consume additional manpower.Therefore,this paper focuses on how to improve the defect detection rate,in order to assist the production line to automatically and accurately complete the defect detection.Aiming at the above problems and considering the current situation of the production line,this paper studies the methods of machine vision and machine learning.At present,the manufacturer has used machine vision and other technologies to complete the image collection and marking in the assembly process.Under the condition of ignoring the marking error,the machine learning algorithm can be used to train the image classification model,which can automatically identify whether the defects exist or not.For this kind of classification task,the support vector machine classification method has full advantages.Under supervised learning,it can automatically learn more complex classification tasks.On the other hand,according to detect the sample's defects,image feature extraction is the premise of machine learning,HOG and LBP feature extraction methods are selected in advance in this paper.Through the research on the application of SVM and image feature extraction methods in vehicle defect detection,a defect detection method based on SVM and HOG is proposed,which can automatically detect the defects in automobile assembly process,and achieve high efficiency and high accuracy of part defect detection.This paper gives a full comparative experiment of the proposed methods,the experimental results show that the method based on SVM and HOG achieves the ideal effect in the application of part defect detection,and the detection accuracy is as high as 94%.At that same time,the method can accurately distinguish the problem of identification error caused by the image position deviation.If apply to actual,the method will greatly save the human cost,save more time and energy for the production process workshop.Thereby improving the production capacity,has the application and popularization value.