| With the vigorous development of science and technology,the popularity of smart phones is getting higher and higher.As one of the fastest updated electronic products,smart phones have the great significance to realize the recycling and reuse of old mobile phones.For the recycling of old mobile phones,the traditional method is to rely on manual completion.There are many problems,such as low price evaluation,large subjectivity,poor convenience,high recycling cost and so on.The old mobile phone intelligent recycling terminal can effectively solve the above problems.The old mobile phone intelligent recycling terminal has the problems of low intelligence,low price valuation,complex operation and so on.In order to solve these shortcomings,a set of old mobile phone appearance detection system is designed in this thesis,and an image acquisition platform is built to carry out the research of determination for the loss and the pricing of old mobile phone.In view of the existing problems and design requirements,a brand and model identification scheme for old mobile phones is put forward.The back of the mobile phone contains the character information other than the mobile phone brand.The edge corner of the old mobile phone contains more contour feat ures of the old mobile phone,and the contour curvature at the edge corner is relatively large.On the basis of using the character recognition algorithm to realize the brand identification of the old mobile phone,the stage template matching algorithm bas ed on curvature is used to identify the old mobile phone model.The first stage is just to match only the peripheral contour information with large curvature.When the model of the mobile phone is not fully determined in the first stage,the weight-based internal contour template matching is carried out in the second stage.Because the internal contour contains less contour information,the measurement factor of the second stage matching effect is multiplied by a weight less than 1,which makes the whole matching algorithm more reasonable.Taking into consideration that the appearance defects of the glass screen have a great influence on the determination for the loss and the pricing of the old mobile phone,this thesis puts forward a scheme for the identification,classification,positioning and edge extraction scheme of the appearance defects of the glass screen.Based on the analysis and definition of the four kinds of appearance defects of the glass screen,the convolutional neural network based on regio n is used to detect and position the glass screen defects.In order to improve the accuracy of the identification of the defect,Generative adversarial networks are added to the convolutional neural network.On the basis of defect identification and positi oning,the detected defects are divided into strip defects and cluster defects.Contour tracking algorithm and fuzzy C-means clustering algorithm are used to extract defect edges,which is convenient for subsequent defect display and manual reexamination.According to the design requirements of the overall scheme,the received information such as brand and model information,glass screen defect information and internal configuration of the old mobile phone are inputted into the determination for the loss and the pricing model of the old mobile phones,and a reasonable recycling price of the old mobile phone is given to complete the determination for the loss and the pricing of the old mobile phones.The feasibility of the overall scheme is verified by experiments. |
