162 - A New Method for Super-resolution Reconstruction...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
A New Method for Super-resolution Reconstruction ABSTRACT In this paper, a super-resolution reconstruction algorithm based on wavelets transform is proposed. Under some special conditions, the blurred low-resolution images can be thought of as the wavelets transform approximate subbands of a high resolution image. Based on the above relationship, we can construct a series of convex sets and then apply the POCS method to recovering high resolution image based on the convex sets. After finite iterative computation, the desired high solution image can be obtained. The experimental results show that the algorithm has good performance in super-resolution reconstruction when the magnification is large enough. Keywords: Super-resolution, POCS, wavelet transform, signal processing I. INTRODUCTION During outer space exploiting, it is necessary to take digital photographs from satellites or space capsules. Due to limitations of digital camera, such as integrated circuit technology, cost etc., desired high-resolution images are not available usually. In order to improve the resolution of pictures, information from low-resolution images can be collected and be used to reconstruct the desired high-resolution image, and this process is called super-resolution. Recently, super-resolution (SR) image reconstruction technology has been one of the most active research areas. Several methods have been proposed for super-resolution image reconstruction 1-6 . In Ref. 1, Tsai and huang gave an approach with Fourier transform. In Ref. 2, Nguyen and Milanfar proposed an efficient method which considers the sampling lattice and employs the wavelet interpolation to reconstruct high resolution image. A regularized method based on multi-channel sampling was proposed in Ref. 3. ML (Maximum Likelihood) methods proposed in Ref. 4, which can be solved by maximum expectation (ME) algorithm. At present, the MAP (Maximum a-Posteriori) method 5, 6 is researched widely. The POCS method is proposed by H. Stark in Ref. 7, which describes an iterative approach to incorporating prior knowledge into the reconstruction process. This POCS method simultaneously solves the restoration and interpolation problem to estimate the super-resolution image. As we known, any signals can be decomposed into two parts after wavelets transform: approximation subband and detail subbands. If some special condition is satisfied, we can consider that the blurred low-resolution images are the wavelet transform approximation subband of a high resolution image. At present, this relationship between high-resolution image and low resolution images was not noticed. Here, we utilize the relationship as the constraints to construct convex sets.
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 4

162 - A New Method for Super-resolution Reconstruction...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online