Grass Field Segmentation, the First Step Toward Player Tracking, Deep Compression, and Content Based Football Image Retrieval

Abstract

AbstractIn this paper, a method is presented which can be used for the segmentation of grass field in video images taken from football matches. Grass field is a green and nearly soft region. Therefore color and texture are two suitable features which can be used to describe it. As HSI color space is more stable against illumination changes in comparison with other color spaces, the hue is selected as color feature.Sub-band images containing high frequency information in horizontal, vertical and diagonal directions that are obtained by applying wavelet transform on image intensity have been used for texture description. Classification of grass and non-grass fields is done using an MLP classifier. The results revealed that the proposed method is able to recognize grass and non-grass samples accurately.KeywordsFootballGrass fieldImage segmentationColorTextureWavelet transformClassification