A Super Resolution Algorithm Based on Iterative Edge-directional Predictions

  • 吳 怡嫻

Student thesis: Master's Thesis

Abstract

This thesis proposes an image super resolution algorithm based on iterative edge-directional predictions The algorithm consists of two major components The first one is an edge-dominate prediction method which applies the gradient between each pair of pixels to obtain the edge areas of the original image The other is an adaptive image enhancement method to compensate the detailed information which is lost in the original image A high-pass filter and an adaptive Gaussian noise are used to analyze and to enhance the texture in the high resolution image PSNR and SSIM criteria are both adopted for the fair evaluation of the performance Experimental results shows that the proposed algorithm achieves 28 564 dB in the average PSNR and 0 9176 in the average SSIM with the lowest computational complexity compared with existing methods
Date of Award2014 Aug 15
Original languageEnglish
SupervisorBin-Da Liu (Supervisor)

Cite this

'