Sensors & Transducers Journal (ISSN 1726- 5479) |
|
Sensors & Transducers Journal 2007 2000-2002 S&T e-Digest Contents
|
||
|
Special Issue: Modern Sensing Technologies
Vol. 90, April 2008, pp. 195-208
Calibration-free Image Sensor Modelling Using Mechanistic Deconvolution
University of New South Wales, Sydney, Australia
E-mail: shen.lim@student.unsw.edu.au
Received: 15 October 2007 /Accepted: 20 February 2008 /Published: 15 April 2008
Abstract: This paper presents a calibration-free approach to modelling image sensors using mechanistic deconvolution, whereby the model is derived using mechanical and electrical properties of the sensor. In this approach, effective focal length is determined using thick lens properties approximated from lens system. The accumulated uncertainties and constraints from sensor properties are utilized with approximated aperture stop position offset to estimate distortion effects, and, eventually to derive corrected image data. This reduces dependency on image data, and, as a result, does not require experimental setup or calibration. An experiment was constructed to evaluate accuracy of model created by the approach, and its robustness to changes in sensor properties without recalibration. The model was compared with a pre-calibrated model using two sensors with different specifications. The model achieved similar accuracy with one-fifth of number of iterations. The approach was also shown to be robust and, in comparison to pre-calibrated model, improved the accuracy significantly.
Keywords: Image sensor, Sensor modelling, Mechanistic deconvolution, Lens distortion, Calibration
Click <here> or title of paper to download the full pages article (1.05 Mb)
|
1999 - 2008 Copyright ©, International Frequency Sensor Association (IFSA). All Rights Reserved.
Home - News - Links - Archives - Tools - Standardization - Patents - Marketplace - Projects - Wish List - Subscribe - Search - e-Shop - Membership
Members Area -Sensors Portal -Training Courses - S&T Digest - For advertisers - Bookstore - Forums - Polls - Sensor Jobs - Site Map