Super resolution based on gradient field
Qiyong Guo, Hongzhi Liu, Wenbin Chen, I-Fan Shen · 2008
In this work, we present a novel method for high resolution image generation from a single low resolution image. The proposed algorithm begins by interpolating the gradient field of low resolution image to obtain one finer gradient field based on local binary pattern feature. Then it recurs to the finer gradient field term and constraint set to construct one reconstruction energy function. By minimizing such energy equation using gradient descent method, a high resolution image can be obtained. We have applied the proposed method to both synthetic data and real image data and comparison results with bicubic interpolation method show that our method is feasible and promising in single image super resolution area.