Research On The Consistency Of Polymer Materials Based On Image Recognition

China has become the world’s largest electrical and electronic products production and consumption country,continuous improvement of productivity,the manufacturing industry especially the household appliance manufacturing industry is developing rapidly,more and more people enjoy the convenience of home appliances.However,the following problems are the quality assurance of polymer materials,which affect the performance of electrical products.At present,the polymer materials determine work can be finished only by experienced experts,by comparing the three types map of polymer materials,analysis the relevant information,and then judges the consistency.It is require high on the professional experience and technical background.Therefore,the task requires the use of efficient way to complete the work.In this paper,a new image recognition algorithm is proposed,which can extract the information of polymer material in real time and accurately.An automatic computer alignment system for the analysis of polymer materials is developed,specific work as follows:1.The recognition and analysis methods of spectral curves are presented,firstly utilize the axis of histogram projection to determine the origin of a coordinate axis,horizontal and vertical coordinates,and then calculate the coordinate scale further calculate the actual physical quantity curve of pixel represents.2.According to the characteristics of the map,we designed two sets of text detection algorithms:(1)The IR map is a regular text in vertical direction,the histogram projection can detect the lines of text quickly and accurately,and then text-line sent to the character recognition module based on template matching.(2)For TG and DSC map,this paper presents a new text detection method for the recognition of map features based on connected region search matching.Firstly,the text region is identified by the connected region analysis.Then,the template matching analysis is carried out in the suspected text region.The highest matching rate class is the result of the matching.3.After image information extraction,it is necessary to compare three types information of the material.In this paper,we propose an algorithm to adjust the threshold of consistency.The algorithm combines the characteristics of three kinds of maps,expert experience and relevant national standards.In the results of 1368 sets of test samples,more than 94%of the results can be determined,while 99%of the determined results are consistent with expert judgment.4.An application system based on B/S architecture for automatic consistency determination is established,the system consists of the following:Front-end WEB server,Back-end MATLAB server and FTP server.Users only need to compare the consistency of the material through the browser.The system provides a friendly man-machine interface and operation mode.The web page covers the "user management at all levels","comparison of parameter settings","consistency judgment" and other functional modules.