In this paper, based on the experimental data, the nonlinear mapping relationship between the apparent in-discharge and the photon number of UV imager in the partial discharge of external insulation equipment is discussed in depth, and the mapping relationship is processed by decision tree judgment and neural network regression respectively. In the processing process, the pollution degree, gain, detection distance, humidity and photon number are taken as the optical signal quantization parameters, and the discharge quantity is taken as the electrical signal quantization parameters. The above two methods are used to process the mapping relationship of the data collected in the test. Through the comparative analysis of the processing results, it is found that the accuracy of decision tree can only reach 46.9%, while the accuracy of neural network regression analysis can reach 96.68%. Therefore, it is reasonable to use neural network to analyze the problem.