LED Chip Identification And Product Detection System Based On Machine Vision

Machine vision is a technology that studies the use of machines instead of human eyes for measurement and control.It has the characteristics of non-contact,high precision,and can work continuously for a long time.The use of machine vision technology in the LED chip placement and product inspection process can solve the low precision and slow speed problem of manual operation.This article analyzes the features and requirements of LED chip identification and product detection process,designs a system of hardware solution,and builds an experimental platform.Based on this experimental platform,the application of machine vision technology in LED manufacturing was studied by using image processing and deep learning algorithm.The main contents of the thesis include:For the LED chip placement process,a precision placement technology based on visual compensation is proposed.By positioning the LED chip before placement,the rotation angle offset is calculated and the mounting angle is adjusted to achieve accurate placement.Firstly,the positive and negative polarities of the LED chip are identified according to the difference in the area of the metal pin area of the LED chip;then the pose angle of the LED chip is identified.And for identifying the pose angle of the LED chip,three methods of identification-Hough transform,morphological connectivity detection and corner detection are used.Through the test data,the accuracy of the above methods and their respective advantages are analyzed.For the detection process of LED products,the detection of LED products based on deep learning was proposed,which improved the poor robustness of traditional image processing algorithms.Under the deep learning framework Caffe,SSD target detection algorithm is used to construct a convolutional neural network and use data sets to train the network.And a well-learned training model is obtained.Using this target detection model,each LED chip in the LED product is identified and extracted,and the perimeter,area,posture are determined one by one,and then determine the LED product is qualified or not.The experimental results show that the use of deep learning algorithm LED product detection system has the advantages of high recognition rate and strong robustness.In the Visual Studio 2013 development environment,combined with OpenCV image processing algorithm library and Caffe deep learning framework,LED chip identification and product detection algorithms are implemented,the overall software framework of the system is built,and designed and written user application software.