Design And Implementation Of Defect Detection Technology For Capacitor Nameplates Based On Machine Vision

With the development of modern science and technology,capacitors are widely used in all kinds of equipment.Because of its superior electromagnetic performance,capacitors become the core components of many high-end electronic devices,and the market is in short supply.Its large-scale production capacity has become the symbol of the development level of a country's electronic industry.The capacitor is produced after the need to go through strict quality inspection to sell,and artificial vision traditional detection methods are time-consuming,costly and often Its loopholes appeared one after another.The automatic capacitor nameplate detection system can solve the above problems.For many defects in the recognition of general artificial visual capacitor nameplate,a full automatic image recognition test system is adopted to collect multi class algorithm.The task of this paper includes the following aspects:firstly,establish the image acquisition module,the surface image of capacitor using CCD camera and FPGA control circuit;then,using wavelet image to obtain the image compression transmission to the following module compression algorithm;and then,in order to avoid the image contains noise or sometimes have brightness the lack of use of bilateral filtering and Retinex image enhancement algorithm to realize image preprocessing;secondly,using the SLIC super pixel segmentation algorithm and fuzzy C means algorithm to divide the capacitor nameplate effective surface information;finally,the use of Caffe deep learning framework and lenet-5 network structure,image text recognition with high accuracy the.Experiments show that the accuracy rate of system segmentation and recognition in this paper is over 90%,which can meet the needs of enterprise naming plate automatic definition test for capacitor inspection.