Image Interpolation Based on Adaptive Neuro-Fuzzy Inference System

Resolution which is defined as the number of pixels per inch is an important factor in determining the quality of images. Due to recent developments, there is a high demand for High Resolution (HR) images. The goal of the methods for increasing the resolution of images, is to produce a high resolution image from a low resolution (LR) image. Preserving image features, including edges is an important factor in image resolution enhancement. Various methods have been introduced in literature to preserve edges. In this paper, the wavelet transform has been used for edge preserving. In proposed method we estimate the four wavelet sub-images of the high resolution image by using adaptive neuro-fuzzy inference system (ANFIS). ANFIS is used for estimating the sub-images due to its ability in using the capabilities of neural networks and fuzzy inference systems (FIS). The method is to use the down-sampled image of low resolution image as input and wavelet sub-images of low resolution image as target outputs, for training four distinct adaptive neuro-fuzzy inference systems. After training, the systems have the ability to estimate the wavelet sub-images of the high resolution image. Finally, by taking inverse wavelet transform from estimated sub-images, the high resolution image is obtained. Qualitative and visual results indicate the superiority of the proposed method to the previous methods.