At present,the automatic recognition task of the pointer meter has become a hot spot of research.The related research mainly focuses on pointer detection,extract line segment characteristics of the pointers,and dial scale layout recognition depends on manual calibration parameters in advance.When the location of the inspection robot moves to the observation point or the deviation of the tilt angle of the PTZ changed,the recognition algorithm may produce large error.In view of the existing defects in the existing methods,This thesis focuses on the automatic recognition algorithm of two types of instruments,the wide scale pointer instrument and the narrow scale pointer instrument,The inspection task of inspection robot and The automatic recognition of dial and pointers are taken as the application background and the target.Firstly,based on the dial scale form and layout law,the dial calibration of the pointer meter dials are located.Then,the default calibration repaired and the instrument scale coordinate system is established.Finally,the pointer is recognized by the relative position relation between the pointer and the scale,and the automatic recognition task of the indicator meter is realized.The main work and innovation of this thesis are as follows:⑴ A feature extraction algorithm for the dial scales: Firstly,an adaptive Canny algorithm is proposed for edge detection with the OTSU algorithm as a priori.Then,the eight domain contour tracking method is used to string the discrete edge points together to obtain the contour.Finally,the contour features are analyzed,and the line segment features and the rectangular features are extracted.⑵ An automatic recognition method for narrow scale pointer instrument and its improvement method: a)Firstly,A line segment extraction algorithm proposed and LSD algorithm used to detect the scale and the line segment of the pointer.Then,the default scale is repaired by a hierarchical nearest neighbor interpolation method,and a complete coordinate system is established.b)Firstly,a line segment pairing based on two Handset Handsets under the constraint of homogeneity is proposed.Then,an improved Ransac ellipse fitting method is designed to obtain a discriminant ellipse.Secondly,an effective default calibration method is studied and a complete scale coordinate system is established.Finally,the recognition pointer of the LSD algorithm is improved.⑶ An automatic recognition method for wide scale pointer instrument and its improvement method: Firstly,the rectangular contour detection algorithm proposed in this thesis is used to identify the dial calibration;Then,the corresponding numbers are obtained by using the scale and the corresponding digital position relation;Secondly,the number is identified and the scale coordinate system is set up;finally,the LSD algorithm is used to locate the pointer.b)Firstly,The dial scale is positioned.Then,the scale is characterized by the radial angle,and the scale coordinate system is set up by the sort.Finally,The LSD algorithm is used to get position of the pointers.⑷ Design of automatic recognition system for indicator instrument: Firstly,the whole process of the software system design is introduced in detail.Then,two software system interfaces are designed with MFC,and each implementation step of the algorithm is obtained and visualized.Finally,the results are analyzed by comparison experiments,and the real-time and effectiveness of the algorithm are verified. |