OpenCV-Based Automatic Detection System for Automobile Meter

Abstract

For the universality of computer vision in the automatic detection of pointer instrument, an OpenCV-based automatic detection system for automobile meter was proposed and realized. The original video image was pre-processed through the filtering and SURF algorithm was used for the registration of template and test instruments to obtain the transformation parameters. Pointer and dial were treated respectively. Through the gray-scale images subtraction, Canny edge extraction and Hough line detection, the straight line where the pointer was obtained rapidly. Then the pointer readings were obtained through functional relation between the point angle and actual value. Experimental results show that this system can be used to conduct the real-time and automatic detection of different types of meters.