Implementation of Color-based Image Segmentation by Clustering Methods

Digital image processing is the fastest growing computer-based technology. It plays a vital role in computer vision and image processing tasks and is generally used for various purposes like medical diagnosis, robotics, remote sensing, industrial inspection etc... Some major operations performed on an image to extract some useful information that corresponds to the key stages in the digital image processing. Digital image processing techniques help in the manipulation of the digital images. While processing the digital image the key phases include pre-processing, enhancement, and display information extraction. The primary objective is to automate multiple tasks together, and the process of image segmentation is one amongst them. The Clustering method is often used for segmenting large scale images for which sometimes preprocessing is required to reduce the volume of data and then other clustering approaches can be applied for better results. This paper experiments the segmentation technique by using the clustering approaches at the same time and considering the parameters like color intensity, the number of clusters required to segment the image and on the basis of mean shift bandwidth. The purpose of this report aims at the implementation and comparative analysis of the results generated using the normalized cut, k means and the mean shift clustering technique. Keywords— Image Segmentation, Normalized Cut, K Means, Mean Shift, Hierarchical Clustering