Automatic Recognition Research Of Square Pointer Meter Based On Machine Vision

A new round of technological revolution and industrial transformation is profoundly affecting traditional industries,which is centered on Internet technology and represented by “Industry 4.0”.With the remarkable achievement of the construction of distribution automation,smart grid and maintenance systems,the efficiency of distribution network operation and inspection has been greatly improved.However,the “manpower-based” maintenance mode is inefficient,and the data mining of the acquisition system is insufficient.It is an effective scheme to introduce the automated inspection robots into the distribution network.Firstly,the system architecture and functional design scheme are established by analyzing the requirements of the on-duty task in the distribution room.The detailed design is carried out from three aspects: 1)system communication,2)monitoring and early warning of distribution status,3)software and hardware system.Then,according to the duty work flow,this paper analyzes the object of the robot vision system and discusses the overall design of the pointer meter recognition system.In view of the shortcomings of square pointer meter recognition and the interference factors of image acquisition environment,a multi-perspective,multi-objective and anti-interference recognition algorithm is proposed.The algorithm consists of three processes: instrument separation,frame positioning and pointer recognition.In the process of instrument separation,the instrument localization algorithm based on the contour feature and “score system” are used to realize the separation and extraction of multi-view and multi-target instruments.In the process of frame positioning,the gradient gray ratio matrix is creatively proposed to approximate the edge of the frame and fix the inner frame position precisely.In the pointer recognition process,the concept of the gray-gradient combined binary image is proposed,and the improved Hough Transform and line segment feature clustering method is used to recognize the pointer.Finally,the tilt correction model based on perspective transformation is used to calculate the pointer angle.The robustness of the algorithm for the interference factors such as shadow,reflection and large angle tilt is the focus and innovation of this paper.Through the analysis of the pointer meter,the shortcomings of the traditional algorithm are improved,and the anti-interference ability of the algorithm is greatly promoted.After the algorithm implementation and experimental verification with OpenCV and C++,the result shows that the algorithm has high accuracy,stability and anti-interference ability,which meets the accuracy requirements and has high application value.