Resumen

JEEVAN, K. M.  y  KRISHNAKUMAR, S.. An algorithm for wavelet thresholding based image denoising by representing images in hexagonal lattice. J. appl. res. technol [online]. 2018, vol.16, n.2, pp.103-114. ISSN 2448-6736.

The existing method of representation for digital images is using square shaped picture elements called pixels in a rectangular grid. Processing based on hexagonal grid is a new approach in image processing. It has various advantages like symmetry, higher angular resolution, consistent connectivity and higher sampling efficiency. Image processing applications like rotation, scaling, edge detection, and compression in hexagonal domain liave already been discussed by many researchers. In this paper we propose an image denoising scheme in hexagonal lattice using wavelet thresholding method. For the thresholding of wavelet coefficients, modified NeighShrink thresholding method is applied. In NeighShrink method, sub-optimal universal threshold and identical neighboring window size in all wavelet sub-bands are used. However, in the proposed method, instead of sub-optimal universal threshold, an óptima! threshold is determined for every wavelet sub-band by the Stein's Unbiased Risk Estímate (SURE). Denoising is performed on images represented in rectangular grid as well as hexagonal grid using modified thresholding method for comparison. MSE, PSNR and SSIM are used for the performance analysis. The obtained results confirm that the proposed method gives better results than existing algorithms.

Palabras llave : Wavelet thresholding; Denoising; Hexagonal lattice; Gabor filter; Interpolation.

        · texto en Inglés     · Inglés ( pdf )