Features analysis for tracking players in water polo

摘要

The most desirable property for object detection is its uniqueness. In this paper we propose a method based on pixel classification for feature detection. For particular class of problems related to detection of objects in water, we analyze standard color models and propose YCbCr as the best color model for describing uniqueness of object. We separate luminance from YCbCr color model and analyze relation of Cb and Cr components for different categories of targeted objects (object parts). We employed four different classification methods to test our hypothesis. Experimental results demonstrate the validity and efficiency of the proposed approach.