Improving Spatial Resolution in Exchange of Temporal Resolution In Aliesed Image Sequences

Improving Spatial Resolution in Exchange of Temporal Resolution In Aliesed Image Sequences

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

View Full Document Right Arrow Icon
in: B.K. Ersboll, P. Johansen (eds.), SCIA’99, Proc. 11th Scandinavian Conference on Image Analysis (Kangerlussuaq, Greenland, June 7-11), Pattern Recognition Society of Denmark, Lyngby, 1999, 493-499. Improving Spatial Resolution in Exchange of Temporal Resolution in Aliased Image Sequences Lucas J. van Vliet and Cris L. Luengo Hendriks Pattern Recognition Group of the Department of Applied Physics Delft University of Technology Lorentzweg 1, 2628 CJ Delft, The Netherlands email: [email protected] http://www.ph.tn.tudelft.nl/~lucas Abstract We present a method to improve the spatial resolution of an aliased image sequence in exchange for a lower temporal resolution. The frames are acquired by a vibrating infrared camera, which yields severely undersampled (aliased) image frames of a static scene. The vibration causes a random sub-pixel shift of the scene before sampling. The proposed method first estimates the relative position of all frames after which the periodic-random samples are interpolated on a equidistant grid with a smaller pixel pitch to improve spatial resolution and (partially) eliminate aliasing. The presented method is robust, is completely data driven and has a moderate computational complexity, which permits a real-time implementation in hardware. 1. Introduction A camera employs a lens to produce a (de)magnified version of the scene onto an image sensor. The lens acts as a low-pass filter whose cut-off frequency depends on the lens aperture. Increasing the aperture of the lens increases the light collection efficiency and permits a higher spatial resolution. An array-based sensor in the image plane not only converts the incoming light into electrical charge, but also samples the image. The Nyquist criterion states that the sampling frequency in the image plane should be higher than twice the cut-off frequency. Satisfying this requirement yields a digital image that contains all the information of the bandlimited analog image. Exact reconstruction of this bandlimited analog image from its samples is possible [7,8]. Infrared cameras employing a 2-D array of light sensitive elements require cooling and isolation of the individual pixels to suppress thermal noise. This yields a pixel pitch that is far too large for high-resolution imaging. The individual frames are undersampled. Undersampling yields aliasing, which corrupts the image data irreversibly. However, a set of aliased realizations of the same scene may under restricted circumstances contain all the information of the original bandlimited image. Our IR camera vibrates, which yields an image sequence in which each individual frame is shifted over a random sub-pixel vector. An image sequence of a (pseudo) static scene may contain enough information to increase the spatial resolution without being hampered by the aliasing, which corrupted the individual frames.
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.

This note was uploaded on 04/22/2010 for the course MI IP taught by Professor Vladbalan during the Spring '10 term at Universidad del Rosario.

Page1 / 7

Improving Spatial Resolution in Exchange of Temporal Resolution In Aliesed Image Sequences

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