Effective Use of Low Resolution Images for Super-Resolution Reconstruction

High resolution image reconstruction is a powerful image processing techniques which sought to obtain a high resolution image from observed multiple low resolution images. High resolution image reconstruction is very useful in various areas such as medical imaging, surveillance and satellite imaging. The method which is proposed in Tong and Leung [8] used linear interpolation to build up an algorithm to obtain the relationship between the detail coefficients in wavelet subbands of the high resolution image and the low resolution images. The method gave better results than Tikhonov least-squares method. Later, in Tong and Leung [9] proposed an error correction extension for their method which provided very good results even for noisy cases. However, we note that the algorithms in [8] and [9] did not make use of all the low resolution image data on an equal footing. Thus in this paper, we propose a method which can effectively use all the observed multiple low resolution images to reconstruct a high resolution image. The experimental results show that our proposed method can work better than the method proposed in Tong and Leung [9], Least-squares and Chan et al.´s Algorithm 3 [3] even for noisy cases.