E1-3 - Sedma Nacionalna Konferencija so Me|unarodno U~estvo...

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

View Full Document Right Arrow Icon
PIXEL-LEVEL SELECTIVE ROBUST SUPER-RESOLUTION Z. Ivanovski 1 , T. Kartalov 1 , Lj. Panovski 1 , Lina J. Karam 2 1 Faculty of Electrical Engineering, Ss Cyril and Methodius University, Karpos II bb, P.O.B. 574, 1000 Skopje, Republic of Macedonia, mars@etf.ukim.edu.mk 2 Department of Electrical Engineering, Arizona State University, Tempe, AZ 85287-5706, karam@asu.edu Abstract – In this paper a new technique for robust super-resolution (SR) from compressed video is presented. The proposed method exploits the differences between the low resolution images at pixel level in order to determine the usability of every pixel in the low resolution images for SR enhancement. Only the pixels from the low resolution images that are determined to be usable are included in the L 2 norm minimization procedure. The results obtained with real video sequences demonstrate superior quality of the resulting enhanced image in the presence of outliers and same quality without outliers when compared to existing L 2 norm minimization techniques. At the same time, the proposed scheme produces sharper images as compared to L 1 norm minimization techniques. Keywords super-resolution, compressed video, compression artifacts, L 2 norm minimization 1. INTRODUCTION The super-resolution (SR) reconstruction problem has been of interest to the DSP community for the past twenty years. During this period a large number of techniques and algorithms have been proposed in the literature. Most of the techniques assume that low- resolution images result from warping, blurring and resampling of high-resolution images. The goal is to find the high resolution image which, when warped, blurred and resampled according to the imaging model, predicts the low-resolution images well. For effective SR accurate motion estimation is essential. Inaccurate motion estimation can introduce visible noise and artifacts, thus degrading the image instead of enhancing it. Scene changes and occlusion can also produce artifacts, which result in poor quality image. In order to deal with the problem of outliers different techniques have been proposed. Farsiu et al. [1] propose a SR scheme based on L 1 norm minimization, showing that the steady state solution for a blurred high-resolution image is the median of the pixels in the corresponding low- resolution images. Recently, the SR reconstruction problem has been extended to the area of compressed video. In order to use compressed video for SR, the problem of compression artifacts should be considered. A number of techniques have been proposed in the literature addressing the issue of simultaneous SR and compression artifacts elimination. Most of them are using special regularization terms to eliminate the compression artifacts in the estimated high-resolution image. This paper presents a new SR scheme that exploits
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

E1-3 - Sedma Nacionalna Konferencija so Me|unarodno U~estvo...

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