Power Equipment Identification And Condition Monitoring In Complicated Background

Image recognition processing technology plays an important role in daily life and industrial production.Monitoring of power equipment in the power industry is also inseparable from the image processing technology.This article is the use of image processing technology for substation inspection robot scene image collected cantilever switch target extraction and state identification.First of all,through the function of the cantilever switch and the on-site state,the paper analyzes the characteristics of the identification target and the basis of the state identification.At the same time,according to the visual characteristics of the inspection robots,the image is processed by means of mean filtering,morphological processing,adaptive binarization and target area interception.And use SVM algorithm and template matching algorithm to identify the image,analyze the characteristics of the results,and explore complex image processing schemes.According to the characteristics of the cantilever switch,the double circular ring contact with obvious characteristics is reconstructed,the double ring model is constructed and the mathematical model is inferred.For each ring body,through the distance between the random point pair the shape distribution curve of the torus is distributed and the voxel is judged as a torus according to the shape distribution curve.For the two-dimensional image of the planar projection of the double ring model,the concept of characteristic chord is introduced and the mathematical description of the general ellipse is given based on the concept of chord.In this paper,an improved algorithm of recognizing elliptical shape using random Hough transform under the action of attached chord transform is used in the field image recognition of cantilever knife.According to the mathematical model,the center of the ellipse is extracted and detected,then the center of the ellipse is fitted to determine the target area.After the target area of the double ring model is extracted,the judgment of the degree of aggregation of the black pixels in the lower part of the extraction section is performed to judge the state of the cantilever switch.In order to make the recognition effect of Hough transform under the characteristic chord effect clear and distinct,it is convenient to display the recognition result and to make the algorithm more operable.Based on the MATLAB GUI,the interface is written.At the same time,this algorithm can be compared with the SVM algorithm and the recognition result of the template matching algorithm.