Back close

Hyperspectral image classification improved with ELRMA denoising

Publication Type : Journal Article

Publisher : International Journal of Control Theory and Applications.

Source : International Journal of Control Theory and Applications, Volume 9, Number 10, p.4603-4609 (2016)

Url :

Campus : Bengaluru, Coimbatore

School : Department of Computer Science and Engineering, School of Engineering

Center : Computational Engineering and Networking

Department : Computer Science, Electronics and Communication

Year : 2016

Abstract : In this paper, scope of improvement of hyperspectral images with a preprocessing technique is studied using various classification methods. Hyperspectral images are of great scope in exploration as it provides wider and precise information. Reflectance of a hyperspectral image contains spectral information of pixels as well as spatial information. Hyperspectral images have wide range of applications in diverse fields of remote sensing such as geology, oil spill detection, land cover classification, mineral detection, bio mass detection, urban planning and forest study. Since hyperspectral images are subjected to noise, denoising using enhanced low rank matrix approximation(ELRMA) is applied as a preprocessing technique. Low rank matrix approximation(LRMA) is enhanced using a non-convex regularization treating it as a convex optimization problem and it is applied to hyperspectral images. Using ELRMA technique denoising of hyperspectral images are done effectively and the improvement is analyzed using subspace pursuit algorithm, GURLS and random forest classification methods. © International Science Press.

Cite this Research Publication : G. Swamynadhan, Dr. Nidhin Prabhakar T. V., and Dr. Soman K. P., “Hyperspectral image classification improved with ELRMA denoising”, International Journal of Control Theory and Applications, vol. 9, pp. 4603-4609, 2016.

Admissions Apply Now