| Abstract: | One of the essential tasks in a wide variety of applications and image processing domains is the need of automatic segmentation. The image segmentation divides an image in to homogeneous regions mainly to locate objects and boundaries such as curves, lines, regions etc. The aim of image segmentation is to extract significant features of image, thus understanding, interpretation, description, analysis of the image scene becomes easy and useful for the machine to perform future tasks. Region segmentation divides images into regions based on pixel intensities, patterns, locations, local shapes and textures or combinations. The region segmentation mainly depends on how effectively the segmentation scheme captures the local attributes. This paper captured the local information by converting the grey level image in to a uniform local binary pattern (ULBP) image. The advantage of this scheme is it only contains the significant information and all non significant local information is represented under a single label. Histogram equalization is used to enhance the local contrast. A morphological treatment is given to this, to smooth the regions and to close narrow gaps between two structures without growing the size of the structures. The Otsu’s threshold is used for the final segmentation. The experimental results on four different databases demonstrate the success of the proposed method, compared to many other methods.
|