10.1.1.90.4387 - Effect of Measurement Precision on...

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

View Full Document Right Arrow Icon
Effect of Measurement Precision on Super-resolution Image Reconstruction Sally L. Wood, Gao Yang Electrical Engineering Department Santa Clara University, Santa Clara, California, 95053 swood@scu.edu , gyang1@scu.edu Marc P. Christensen, Dinesh Rajan Electrical Engineering Department Southern Methodist University Dallas, Texas 75275 mpc@engr.smu.edu , rajand@engr.smu.edu Abstract: Low-resolution 8-bit and 14-bit images were used to reconstruct high-resolution images. Performance is measured using standard test charts and compared with analytical predictions and simulations to determine the quantization effects on the reconstructions. OCIS codes: (110.0110) Imaging systems, (100.6640) Super-resolution 1. Introduction Computing high resolution (HR) images from multiple offset low resolution (LR) images has been analyzed by a number of investigators for a variety of applications [1-7]. Implementations often use readily available web cameras and rely on sequential acquisition of multiple shifted LR images. Designs for simultaneous acquisition of multiple LR images include fixed arrays of imagers [8-9] or a dynamically controlled array of sub-imagers [10]. Analytical results and simulations show that increasing the resolution improvement factor also increases the average noise per pixel of the HR reconstruction when the input noise for the LR images is held constant [4,5]. Noise may be due to random uncorrelated measurement noise, Poisson distributed counting statistics, and quantization noise based on the number of bits per pixel. Simulations often use unrealistic assumptions or constraints, so algorithms must be tested on real image data. In this work we present reconstructions at multiple resolutions based on data gathered with differing quantization. The goal is to quantify the effects of quantization error on the resultant reconstructed images. 2. Performance Using 8-bit and 14-bit LR Images The objective of the super-resolution image reconstruction is to computationally create a sampled image of f(x,y) at any desired high resolution, where f(x,y) is the continuous image on the detector plane. The upper limit on the desired resolution is determined by the optics. In practice, for a perfectly focused image, the actual resolution of a discrete sampled image is limited by p(x,y), the response of the individual detectors. The function defining the response of a detector at any position (x
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 / 3

10.1.1.90.4387 - Effect of Measurement Precision on...

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