Automatic Grid Segmentation of Populated Chessboard Taken at a Lower Angle View

Abstract

Segmentation of the chessboard grid is a crucial step in coordinate extraction for chess video annotation, content-based indexing of chess videos and for content analysis in prototyping vision systems such as a chess robot. This paper deals with the segmentation of grid elements from a populated chessboard taken at a lower angle view. The proposed approach couples the line-based grid detection method with domain knowledge of the chessboard to achieve improved accuracy and reliability.