A fast component labeling and description algorithm for RoboCup middle-Size League

For RoboCup middle-size league, vision processing have its particular characteristics. The requirement of real time and multi-channel image segmentation processing make the algorithm of component encode consume less processing time. In this paper, the new component encode algorithm, which is based on run-length encode, uses a new run structure and other assistant data structure to reduce the memory spaces, these structures make run movement and comparison more effective. This paper also proposes the methods of run movement, run comparison and index combination. Experiment show the proposed algorithm remarkably reduces processing time, run comparison and index combination. The proposed algorithm is very suitable for the vision processing in RoboCup middle-size league.