Multivariate Shape analysis using modified Otsu’s method for High-Resolution Remote-Sensing Image Segmentation

Abstract

The image segmentation is mentioned to as a vital practices of image processing. It is the procedure of separating or apportioning an image into portions, named segments. It is typically suitable for applications alike image compression or object identification, since for them, it is incompetent to practice the entire image. Therefore, image segmentation is employed to divide the portions from an image for additional handling. Various image segmentation methods are availed, which divide the image into numerous portions according with definite image topographies such as pixel intensity, color, texture, etc. In this study, the protuberant approaches such as local homogeneity analysis, openings and closings with reconstruction criteria and modified Otsu’s method have been utilized for image segmentation. While comparing the obtained results of local homogeneity analysis and openings and closings with reconstruction criteria, the modified Otsu’s method offers adequate outcomes for the given images with pertinent histograms. The simulation studies reveal that the modified Otsu’s method has the benefits of real time and assured anti-noise skills, the contaminant can be found accurately and made the segmentation process ease and consistent.