Monocular Vision Measurement System Based on ARM Microprocessor

Abstract

The target location problem of multi-robot cooperation and service robot tracking is discussed.A monocular vision measurement system with the ARM9 microprocessor as the main processing module and the CMOS camera as the image capturing module is presented. The system design includes two parts,i.e.,object recognition and positioning.A pure color block is served as the goal object, which can be detected by searching the color block in the captured image.The color image threshold segmentation method based on RGB vector is used to implement the image segmentation.The P4P method is employed to achieve the distance measurement.The intrinsic camera parameter calibration is performed with the Matlab toolbox and the RANSAC method is utilized to find the edge lines of the color block.The experiments show that the system is accurate and robust.