Research On Method Of Insertion And Detection For Vehicle Fuse Box Based On Machine Vision

The vehicle fuse box is one of the important electronic control units in the car carrier.Its main role is to ensure the safe operation of the vehicle circuit and to prevent the vehicle carrier circuit from damaging the electronic equipment in the car when an abnormal current occurs.It consists of an inserting box and vehicle fuse chips.According to different requirements,fuse chips are inserted and arranged in an inserting box.Currently,the majority of fuse boxes are inserted and inspected manually.Since different types of fuse chips have strong similarity,manual inspection for fuse boxes requires more labour.Still,it is a very complicated task for humans also due to its monotonous nature of work,which leads to a loss of concentration which may prove to be lethal afterword.There often is inverse relationship between accurate rate of manual method and work efficiency.The error rate increase significantly if a human operator just gets a short time to insert and check a fuse box.Thus,it is of practical significance to study a method that can be used to realize automatic insertion and detection of vehicle fuse boxes.The production of vehicle insurance boxes faced the problem of how to recognize vehicle insurance tablets accurately and insert rapidly.Aiming to solve the problem,we propose a method identifying and inserting vehicle insurance boxes based on color vector clustering.This method uses the industrial CCD camera to obtain image information of vehicle insurance tablets.We adopt the average background method to eliminate the image background and stain repair technology to enhance the image.We regard the center color vector of different kinds of vehicle insurance tablets as the according of features extraction,which obtained from image information.Then,we input the feature vector into support vector machine so that the classification and identification of different types of insurance can be achieved.Based on the classification results,four-axis robot,SCARA is controlled to complete the accurate insertion.Based on many verification experiments,the method of recognizing toward the 9 types of vehicle insurance tablets can obtain accuracy rate of 99.7%.When the average plug cycle is 1s,the correct rate is above 95.6%.Aiming at fuse box detection,we propose a method detecting fuse boxes based on polar coordinate feature matrix(PCFM).This method employs a CCD camera to acquire the image of fuse boxes,employ histogram of oriented gradient(HOG)feature to characterize the ID code of vehicle chips and combine SVMs to accomplish the location and recognition of fuse chips.The Arrangement structure of fuse chips in fuse box is described by PCFM,which is calculated based on the results of location and recognition.Finally,vehicle fuse box detection is realized based on PCFM similarity.The experiment shows that the accuracy for vehicle fuse box detection,based on PFM similarity,reached 97.6%.