Image Restoration Based on PLS and Wavelet Bi-cubic Interpolation
Tian Zho · Journal of Yangtze University · 2013
For human face image restoration problem,a novel super-resolution algorithm is put forward based on Partial Least Squares(PLS)regression and a wavelet bi-cubic ratio interpolation algorithm.The low resolution image is decomposed into three high-frequency images and a low-frequency image by wavelet transform.Three high-frequency images are interpolated before reconstruction by wavelet inverse transform.At the same time,the original low resolution image is interpolated by bi-cubic interpolation.The interpolated image and reconstructed image by wavelet transform are fused.The characteristics are abstracted from fused images as trained samples.The high resolution image is restored according to partial least squares regression.It is shown from experimental results that the super-resolution restoration algorithm based on partial least squares and wavelet bi-cubic ratio interpolation algorithm is better than the other traditional algorithms.