Copper Electrolysis Short Circuit Detection Of Infrared Thermal Image

In the process of copper electrolysis, the uneven current distribution of the anode and cathode plates can lead to the formation of copper particles on the cathode plate. At this time the cathode plate will be short circuit and the temperature of cathode plate will be rise. In this paper, the thermal infrared image is used to detect the short circuit plate, and it is more timely and accurate than the traditional method.The innovation and key work of this paper can be summarized as follows:(1) Image preprocessing section:First, in view of the phenomenon that the infrared image have barrel distortion, this paper calibrate the image by geometric calibration.Second, in view of the need of obtain the temperature, this paper calibrate the temperature of infrared camera. Third, several segmentation algorithms are compared and the segmentation scheme is given.(2) This paper presents a sample collection scheme, and collectes 5000 circuit plate and normal plate sample. In order to facilitate the text, the short circuit plate and the normal plate are named as negative samples and positive samples. After statistical analysis, the paper gives the selection criteria of the middle column of the sample, and the optimum width of the sample.(3) Image of electrolytic tank which covered with thermal insulation cloth will makes negative sample temperature information lost, at this point it is difficult to use absolute temperature information to detect negative samples, so it is necessary to extract more stable features. This paper presents two kinds of features for the classification of positive and negative samples:first, the differential LBP feature description operator is proposed based on the local binary pattern; second, the PCA feature of image pixel value ordered is proposed based on the principal component analysis. These two features have achieved good results in the classification and recognition of positive and negative samples.(4) This paper using Statistical prior knowledge and low dimensional feature to improve the search rate of negative samples, eventually make the search speed increased by about 7 times.