Alignment invariant image comparison implemented on the GPU
Highquest, Johannesburg
Nineteenth Annual Symposium of the Pattern Recognition Association of South Africa (2008) Volume: 1, Publisher: Pattern Recognition Association of South Africa, Pages: 141-145
@article{roosalignment,
title={Alignment invariant image comparison implemented on the GPU},
author={Roos, H. and Roodt, Y. and Clarke, W.A.},
year={2008}
}
This paper proposes a GPU implemented algorithm to determine the differences between two binary images using Distance Transformations. These differences are invariant to slight rotation and offsets, making the technique ideal for comparisons between images that are not perfectly aligned. The parallel processing capabilities of the GPU allows for faster implementation than on traditional desktop processors. In order to take full advantage of this all aspects of the algorithm was implemented on the GPU.
March 16, 2011 by hgpu
