Research On Automatic Recognition Of Pointer Dial Of Inspection Robot Based On Computer Vision

With the development of the times,electric power plays a pivotal role in national economic development and people's life.The stable operation of the power grid system has a great impact on society.In our national power system,substations are the most core part.To ensure the stable operation of substation equipment in the substation,it is necessary to conduct regular inspections on the substation equipment.At present,the operation status of the substation equipment is judged mainly by reading the meter in the substation.Nowadays,most of the substation equipment is monitored by manual inspection of reading meters,and the inspection method is not only costly but also has low reading accuracy.Although there are many studies on the application of computer vision technology to the automatic recognition of the dial to replace manual inspection,there are many constraints in the recognition algorithm to have a good recognition effect and it is difficult to adapt to the complex and variable substation environment without considering the detection efficiency.This paper revolves around the existing shortcomings and limitations of the automatic recognition of the pointer dial.It systematically researches the dial positioning algorithm,the dial reading area localization algorithm,the pointer positioning method and so on.And it realize an intelligent reading method based on computer vision for pointer dial of inspection robot in substation environment The core work and research contents of this paper are as follows:(1)In aspect of dial area location,this paper proposes an improved pointer dial area location algorithm based on SSD model.In view of the traditional feature point matching algorithm for big images detection will be time consuming and the problem of poor robustness.This paper transforms the pointer-style dial region localization problem into the target detection problem in the deep learning field,and migrates the learning strategy to improve the SSD-based model.And data enhancement method is used to further enhance the dial positioning effect.In this paper,the dial positioning algorithm not only reduces the positioning time,but also improves the adaptability to noise,light change,rotation and other interference factors.(2)In aspect of the segmentation of dial reading area,this paper proposes a multitask template feature point matching algorithm based on SURF algorithm.Aiming at the difference between the template produced by the large outdoor illumination and the detected image,which leads to the problem of inaccurate matching,this paper uses multithreading technology to simultaneously perform feature point detection with the detected image by making multiple templates with large photogrammetry.And the improved bidirectional matching method proposed in this paper is used to obtain the matching points,so as to improve the accuracy of the division of the reading area of the dial and enhance the robustness of the matching algorithm to the large illumination variation.(3)In the aspect of image preprocessing,this paper studies the image preprocessing algorithm suitable for complex working scenes of substation.Through the gray processing,image contrast stretching and filtering decontamination preprocessing techniques,the images with good image quality and convenient segmentation pointers are obtained.(4)In terms of pointer positioning and recognition,this paper proposes a pointer positioning method based on segmented sector regions.Aiming at the problem that the pointer is difficult to segment due to insufficient illumination and complex background interference,this paper proposes a pointer positioning method based on segmented sector based on the characteristics that the dial pointer has been rotating in a certain sector.It locates the pointer by segmenting the pointer active region and selecting the appropriate threshold segmentation algorithm and pointer fitting algorithm.The pointer positioning method proposed in this paper can accurately fit the pointer line and has good robustness.According to the automatic recognition scheme of pointer dial based on computer vision proposed in this paper,a software system of pointer dial recognition based on computer vision is designed and made.Through experimental verification,the recognition rate and accuracy of the scheme have reached the practical requirements and have high stability and detection efficiency,and have certain practical value.