Realtime automatic level bar calibration based on Canny edge detection and weighted least squares method
SHENG Wei1, WANG Qingguo2, ZHU Shanan1
（1.College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China; 2.Institute of Intelligent Systems, School of Electrical Engineering, Johannesburg 2000, South Africa）
 

Abstract: Aiming at the automatic calibration problem of the bubble level,the edge detection algorithm,the least square method and the contour tracking algorithm in the field of machine vision were studied. A realtime automatic detection method for the level bar based on Canny edge detection and weighted least square method was proposed to improve the accuracy and efficiency of the bubble detection. The Canny edge extraction algorithm for image preprocessing was introduced to deal with the image of the level column taken by the industrial camera to get the edge information of the reference lines and the bubble. The thresholds of Canny edge extraction algorithm was set adaptively to reduce the effects of changes in the light field on detection accuracy. The weighted least squares was employed to model the parallel reference lines and the binary search algorithm was used to search the bubble position, so as to achieve highaccuracy detection of the level bar. The results indicate that the proposed method can accurately and quickly locate the bubble,and can well adapt to the changes of the light.
Key words: image processing; Canny edge detection; weighted least squares; realtime detection; level bar