Super resolution image reconstruction using penalized-spline and phase congruency. (August 2017)

• Record Type: Journal Article
• Authors: Nayak, Rajashree; Patra, Dipti
• Is Part Of: Computers & electrical engineering. Volume 62(2017)
• Page Start: 232
• Page End: 248
• Publication Date: 2017-08
• Subjects: Super resolution -- Image reconstruction -- Structural regularization -- High frequency energy regularization -- Phase congruency -- Penalized-spline interpolation; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2016.10.003
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: Regularization based super resolution image reconstruction. Reconstruction process maintains consistency of structural features. Preserves the complete set of high frequency components. Invariant to illumination and/or intensity variations in images. High quality solution with improved initial image and adaptive learning rate. Graphical abstract: Abstract: Super resolution reconstruction (SRR) of image is an ill-posed inverse problem. Several regularization based SRR (RSRR) methods have been suggested in the literature to obtain a unique and stable solution. In this paper, we propose a new RSRR framework by imposing two regularization constraints such as structural regularization term (SRT) and high frequency energy term (HFET). SRT utilizes the local phase coherence of the image to maintain consistency of structural features in the reconstruction process. HFET enables to preserve the complete set of high frequency components using phase congruency based on 2D Hilbert transform. Gradient descent method optimizes the regularized cost function. The optimization process starts with an improved initial high resolution (HR) image obtained by Penalized-spline interpolation. The iterative process progresses with an adaptive learning rate to yield a high quality HR solution at a faster rate of convergence. Simulation results validate the effectiveness of the proposed regularization scheme.