iNEDI (improved New Edge-Directed Interpolation) Image Zooming Algorithm
- category Application
- package \Com\Tecnick\inedi
- author Nicola Asuni info@tecnick.com
- copyright 2006-2016 Nicola Asuni - Tecnick.com LTD
- license http://www.gnu.org/copyleft/lesser.html GNU-LGPL v3 (see LICENSE.TXT)
- link https://github.com/tecnickcom/inedi
- version 1.6.2
This projects contains a Matlab implementation of the iNEDI (improved New Edge-Directed Interpolation) algorithm to enlarge (zoom) digital images.
This inedi.m function returns an enlarged image by a factor of 2^ZK for both horizontal and vertical directions.
This function implements the technicque described in: N. Asuni, "iNEDI - Tecnica non lineare per l'interpolazione di immagini," Master's thesis - University of Cagliari, 2007. Thesis [ITA]: http://www.tecnick.com/pagefiles/appunti/iNEDI_tesi_Nicola_Asuni.pdf Presentation [ITA]: http://www.tecnick.com/pagefiles/appunti/iNEDI_presentazione_Nicola_Asuni.pdf
[EIM] = inedi(IM) [EIM] = inedi(IM, ZK) [EIM] = inedi(IM, ZK, MT) [EIM] = inedi(IM, ZK, MT, ML) [EIM] = inedi(IM, ZK, MT, ML, BT) [EIM] = inedi(IM, ZK, MT, ML, BT, BS) [EIM] = inedi(IM, ZK, MT, ML, BT, BS, SZ) [EIM] = inedi(IM, ZK, MT, ML, BT, BS, SZ, VR)
IM : source image ZK : power of the zoom factor (default = 1) the image enlargement on vertical and horizontal direction is 2^ZK; the final image size will be (SIZE * 2^ZK) - (2^ZK - 1) MT : maximum radius of the local circular window (default = 9) ML : minimum radius of the local circular window (default = 3) BT : brightness tollerance (default = 32) Y points are choosen from the ones inside a circular area around the pixel to be interpolated. These pixels must have a brightness between a range defined by the minimum and maximum brightness values of the 4 neighbors pixels of the pixel to interpolate. This range is extended by +/- BT value. BS : maximum brigthess difference for smooth areas (default = 4) This parameter is compared to the difference between the maximum and minimum brightness value of 4 neighbors pixels to discriminate between smooth areas (bicubic interpolation) and edge areas (iNEDI interpolation). SZ : number of image bits per layer (default = 8) VR : verbose mode (default = false) if true prints some information during calculation
EIM : enlarged image
Please check the inediexample.m file on how to use this function.
This implementation is not intended to be used in a production environment. The main purpose of this script is to clearly show how this technique works. Better performaces could be obtained using a compiled version or rewriting this technique using a low-level programming language.
iNEDI, NEDI, image, zooming, magnification, upsizing, resampling, resolution enhancement, interpolation, super-resolution, covariance-based adaptation, geometric regularity, matlab, octave.