Summary
the 2014 International Symposium on Nonlinear Theory and its Applications
2014
Session Number:C3L-A
Session:
Number:C3L-A1
Directive Cycle-Spinning Cellular Neural Network for High Frequency Subband Interpolation
Kosuke Hosaka, Tsuyoshi Otake, Hisashi Aomori, Masatoshi Sato, Mamoru Tanaka,
pp.565-568
Publication Date:2014/9/14
Online ISSN:2188-5079
Summary:
In this paper, we propose a novel image resolution enhancement technique based on the interpolation of high-frequency components of discrete wavelet transform (DWT). The low-resolution (LR) image whose resolution is to be enhanced is decomposed into four subbands by DWT on the basis of lifting method. Then only the high frequency subbands are interpolated with magnified factor α by using cycle spinning cellular neural network (CS-CNN). The CS-CNN is particularly well-suited for solving nonlinear optimization problems defined in space such as image processing tasks. In our algorithm, a directive architecture using a CS-CNN is developed to prevent the unnecessary smoothing of image detail. While a discrete-time cellular neural network (DTCNN) transforms all high-frequency subbands of LR image into coefficients to predict the original subbands of high-resolution (HR) image using the A-template, the directive cycle spinning method is applied to estimate the optimal coefficients from individual outputs of the DTCNN as above. Experimental results indicate that the proposed method produces better results than the conventional image resolution enhancement methods.