Automatic Optical Inspection Detection Of Explosive Dangerous Goods

With the development of technology,the rapid e-commerce education has been the popularity of e-commerce industry has been rapid progress.Automatic Optical Inspection(AOI)is widely applied into lots of products quality testing field.Among them,quality testing of military weapons is quite distinctive,the research on the on-line detection with AOI technique of explosive dangerous goods is the important and difficult problem,on the one hand it is required recognize the character encoding on the warhead,on the other hand it needed to identify all nonconforming products though length detection,bomb gap detection and concentricity detection.So far the detection algorithm is not satisfied the requirement because of the slow speed and incomplete detection.This thesis research on the detection of explosive dangerous goods,on the basis of the original detection add the bomb gap detection,meanwhile raise the speed of detection.Before the detection,build the optical platform in order to get homogeneous light and conveniently obtain completely character encoding information.The main work of this paper is as follows:Firstly,introduce the equipment of explosive dangerous goods system,focused on two parts including character encoding detection devices and edge detection devices.Design imaging system according to the detection needs,build optic platform with homogeneous light.Secondly,extract the useful character encoding information from warhead of explosive dangerous goods.In order to ensure the integrity of useful character encoding,the first thing is get exact position of circular mark template,and take it as the center to extract the effective area contain character encoding.Next segmentation the area,on the basis of finding effective area,comparing with the traditional mesh stretching method,put forward schemes of improvement which is based on each of character encoding instead of the whole area,and then return the character encoding to normal.Finally,put a detection algorithm for edge detection of the explosive dangerous goods,compare with many edge detection and then proposed an improved sub-pixel edge detection algorithm based on ZOM moment.Furthermore,propose an improvedHough transformation axis detection for the detection of length and concentricity.Using NVIDIA CUDA technology to raise the speed of detection algorithm.The characteristic of this dissertation includes the following aspects: research and realize the automatic AOI detection system,the recognition rate is 97.2%,the detection of length and concentricity is 100%,the detection accuracy of bomb gap is 95%,all detect time is 0.59 seconds.