Research On Super-Resolution Of Infrared Image Of Distribution Equipment Based On Two-Step Total Variational Method

Under the background of the rapid progress of information technology,the power system introduces the Internet of things technology and uses intelligent means to monitor and manage the production and transmission process of power.To realize the power Internet of things technology,a large number of sensors need to be installed on the power equipment.Collecting infrared images through infrared sensors is an important means of condition monitoring of power equipment.However,due to the impact of cost and data transmission problems,the infrared images will have the problems of low resolution and insufficient definition.To solve this problem,this paper combines the image degradation model and compressed sensing theory to study the super-resolution method of infrared image of distribution equipment base d on two-step total variation.This research has important theoretical significance and application value.The main work and achievements of this paper are as follows:1)Under the background of power Internet of things,aiming at the problem of low resolution of infrared images of power equipment collected in the process of security monitoring of distribution network,a two-step full variational infrared image super-resolution model is established.In the basic model of compressed sensing,the fuzzy matrix is introduced according to the principle of image degradation to achieve the effect of deblurring the infrared images of power equipment while super-resolution.In addition,in order to solve the problem of "block effect" caused by block reconstruction,the full variational regularization term is introduced into the model to eliminate the boundary noise of the reconstructed image.2)In order to obtain the unknown fuzzy matrix in the super-resolution model in this paper,a fuzzy kernel estimation model based on image gradient a priori is proposed.The change law of significant edge intensity before and after image blur is analyzed,and a new norm is defined,which is introduced into the fuzzy kernel estimation algorithm under the compressed sensing framework as a regularization prior.In the iterative process,the gradient image is constrained,and the significant edge is selected and enhanced to make the intermediate latent image generated in the iteration close to the clear image,so as to estimate the fuzzy kernel according to the low resolution image.3)A two-step total variational image super-resolution reconstruction algorithm is proposed.The svd-k gradient algorithm is used to minimize the sparse kernel of the infrared image,and the svd-k gradient algorithm is used to generate the fuzzy image quickly.The proposed algorithm can effectively deal with the compressed sensing super-resolution model with sparse dictionary,and breaks through the limitation of reconstruction performance caused by the lack of sparsity in the signal transform domain in the regularized compressed sensing super-resolution method.4)In order to verify the application effect of the proposed super-resolution algorithm in power grid,this paper uses super-pixel segmentation,image registration and recognition methods to extract,evaluate and diagnose the abnormal area of power equipment for the high-resolution images obtained by different reconstruction algorithms.Experiments show that the infrared image super-resolution algorithm proposed in this paper can effectively extract the fault area of power equipment and improve the accuracy of equipment defect diagnosis.