Study On The Detection Accuracy Of Static Images In Economical Image Sensors

As the ‘Industry 4.0' and the ‘Made in China 2025' plan are made,our manufacturing industry has the deeper transformation and development,and improving the degree of manufacturing automation is an important goal,especially in the influence of Internet+,the application of machine vision is more and more wide in industrial manufacturing,defense,transportation,natural disasters prevention,medical and other industries,the robot automation technology has become a research hot.And machine vision is one of the core technologies of robotic automation,how to improve the speed and inspection reliability of machine vision,meet the higher precision requirement have become the attention of more and more scholars and manufacturing industries.The static image detection system of economical image sensors studied in this paper is based on the image processing technology,by optimizing algorithms of image feature extraction and image recognition and classification,it has improved the static image detection accuracy of economical type image sensors,and experiments are performed by using a simulation tool.The test system can effectively improve the image accuracy and reduce the error recognition rate,so it avoids the mistake and missing packing of the assembly equipment to increase the productivity of manufacturing enterprises.According to the principle of image processing technologies,this paper focuses on the research of the following several partial contents:Image preprocessing technology.This paper make a pretreatment to images collected by the image acquisition system,it not only uses the traditional spatial mean and median filtering,but make the morphology processing to apply to the image preprocessing,and it effectively reduces the influence of noise factors to the precision of detection system.Image binarization technology.This paper introduces theoretical principles of the traditional histogram analysis and the Ostu algorithm,and makes a experimental analysis,eventually it adopts the Ostu algorithm as the image binarization threshold method of the detection system.Edge detection technology.This paper mainly introduced three kind commonly used edge detection algorithms: the Sobel algorithm,the LOG algorithm and the Canny algorithm,and makes a experiment to contrast the treatment effect of three methods,finally adopts the Canny algorithm as the image edge detection algorithm.Feature extraction,image recognition and classification technology.This article focuses on these two aspects of content.It makes Hu invariant moments of images as images features,and uses the BP neural network to recognize and classify images.Then it makes an experiment through the MATLAB simulation tool,adds the image area area as the image characteristics by experimental analysis,so it meets the precision requirements of the detection system.