Title: Modified silhouette based segmentation outperforming in the presence of intensity inhomogeneity in the hyperspectral images

Authors: Kriti; Urvashi Garg

Addresses: Department of Computer Science and Engineering, Chandigarh University, Mohali, Gharuan, 140413, India ' Department of Computer Science and Engineering, Chandigarh University, Mohali, Gharuan, 140413, India

Abstract: In the real world the intensity inhomogeneity occurs presenting considerable challenges in the field of segmentation of images. In the arena of computer vision and processing of images, the method of level resolute takes an important place. For the conventional level resolute formulations, the indiscretions are established by the function of the level resolute during its progression causing the mathematical error and ultimately destroying the evolution constancy. Hence, a statistical remedy known as reestablishment was applied usually to substitute the besmirched function of level resolute. In the prototypical, the concentration of the confined image is labelled by Gaussian disseminations. It yields a novel type of evolution with level resolute in which the effect of distance regularisation disregards the necessity for reestablishment avoiding its tempted numerical errors. Hence, in the paper a subsequent multiplicative model is presented that considers the inhomogeneity in intensity likewise considering the astronomy of imaging in a diversity of modalities.

Keywords: segmentation of image; intensity inhomogeneity; level resolute method; bias correction; reestablishment; HSI; hyperspectral image; multiplicative model; remoteness regularisation; ascent drift; Gaussian distribution.

DOI: 10.1504/IJIEI.2021.118271

International Journal of Intelligent Engineering Informatics, 2021 Vol.9 No.3, pp.260 - 275

Received: 27 Sep 2020
Accepted: 20 Mar 2021

Published online: 12 Oct 2021 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article