| With the development of Chinese economy, China has become the largest production and consumption country of electrical products in the world. Non-metal materials are key parts of electrical products, of which the quality control is particularly important for product safety, reliability. By comparing the analysis maps of sample and standard library, including differential scanning calorimetry (DSC), thermal gravimetric analysis (TG) and infrared spectroscopy (IR), the composition of non-metal materials can be detected and identified. At present, maps matching works can only be done by manual work, which has high requirements for the technical ability and professional experience of the engineers, and also has the disadvantages of higher repeatability and lower efficiency. Therefore, there is an urgent need to establish an effective way to replace the manual work. This paper develops an analysis map automatic comparison system, which is based on image processing techniques. The main contents of the paper are as follows:(1) Combined with the features of analysis maps, the graphic data in the analysis maps is recognized based on the digital images processing technology, which includes maps’coordinate axis, scale and the data curve. By a large number of identification of analysis maps, the difference of average wave numbers and average error rate are respectively 7.1 and 0.183%, reaching the desired effect。(2) Use the feature extraction algorithms such as uniform grid, directional line element, the standard LBP and an improved LBP to complete the feature extraction of characters. Then use template matching and SVM algorithms to classify the results of the feature extraction. The scheme which combines the improved LBP and template matching algorithms has a better recognition effect, and the character recognition rate and the rejection rate reach 98.2% and 0.25% respectively.(3) A threshold comparison algorithm is proposed in this paper, which can be used to determine the consistency of the three types of analysis maps. The graphical information and character information previously distinguished are used to determine the consistency of maps. Finally, an integrated automatic material consistency judgment system is established with MATLAB GUI. If the consistency determination accuracy is set to100%, the system recognition rate can reach about 90%. |
