One of the essential tasks in a wide variety of applications and image processing domains is the need of automatic color segmentation. The aim of color image segmentation is to extract significant features of image, thus understanding, interpretation, description, analysis of the image scene becomes easy. The present paper attempted this by deriving an efficient and simple color segmentation approach. The proposed segmentation approach initially extracts the red, green and blue color components separately. On each color component the present paper computes the fundamental texture units that represent edges, flat regions, corners, spots and other significant local attributes by reducing the dimensionality and by achieving rotational invariance. To fill the narrow gaps in between adjacent edges, corners morphological operation is performed. K-means clustering is performed to segment the local regions of individual color components more significantly. Finally the color components are concatenated to generate the final color segmented textures. The quality measures are applied to test the efficacy of the proposed method. The proposed method is implemented on Wang, Oxford flower, MIT-VisTex databases and compared with state-of-art of the existing methods.
展开▼
