Recognition andpositioning technique of Chinese chess based on vision

Abstract

A technological solution was proposed based on vision to identify and locate Chinese chess pieces according to the requirements on chess recognition and accurate positioning of a Chinese chess robot.The pieces were identified using character segmentation on the post-correction image,the Moravec operator to extract the character feature,and the BP network algorithm.The robot achieved the accurate positioning of the pieces by means of changing the image coordinates of the identified pieces to actual chessboard coordinates.Tests show that the chess robot can correctly identify all the pieces by adopting this solution,with the piece location range error of within 1cm,the maximum error percentage of 5.468%,and the minimum error percentage of 0.533%,which meets the requirements of Chinese chess robot positioning.