Pointer Instrument Recognition Algorithm Based on Haar-Like Feature and Polar Expansion
Xinxin Zheng, Xia Chen, Xiao Long Zhou, Xingang Mou · 2018
In order to improve the degree of automation of instrument reading recognition process, this paper proposes a new reading recognition algorithm based on digital image processing technology for a kind of pointer instrument with complex background. Using the transcendental characteristics, image contour information was used to locate instrument, image morphology method was used to eliminate interference in text and Haar-like edge feature was used to extract and filter. Correctly identifying the instrument calibration is the precondition for the accurately reading, project for many times on the basis of the traditional projection method of Polar expansion in the horizontal and vertical direction, analysing projection curved to extract pointer position and calibration, and calculating the reading according to the pixel. The test results show that the algorithm is stable and reliable, which can effectively to remove the background of the instrument, has high real-time and computational accuracy and meets the application requirements of engineering.