A fast SISR technique founded on synthesize high-band frequency and an adaptive hampel stochastic function

Abstract

Theoretically, the conventional image enlarging technique is a mathematical method for building a superior enriched resolution image, which is normally insisted for modern computer vision and image processing application by utilizing only one poor resolution image, which is normally captured from any commercial embedded camera systems. Due to the fast computation, the Single-Image Super-Resolution (SISR) is one of the well-known Super Resolution-Reconstruction (SRR) techniques and the SISR is desired for applying on only one poor resolution image. Therefore, this article aims to present the image enlarged technique founded on SISR algorithm utilizing Hampel stochastic function and high-band frequency synthesizing. Unfortunately, the efficacy of the SISR technique is relied upon three parameters (b, h, k) and it is difficult task for estimating these suitable values of these three parameters for reconstructing the superior enriched resolution image with the optimum Peak Signal-to-Noise Ratio (PSNR). In consideration of deciphering to this problem, the Hampel stochastic function, which is relied upon wholly one parameter (J), instead of three parameters like the conventional function, is comprised into SISR technique. By studying on 14 classic images, which are corrupted by different noise models, in the statically exploratory section, the efficacy of the fast SISR technique closely equal to the conventional SISR technique but the parameter adjustment process of the proposed fast SISR technique (with one parameter) is more simple and fasert than the conventional SISR technique (with three parameters). Because of fast computation in the parameter adjustment process, the proposed fast SISR technique is more suitable for real implementation.