Skip to Main content Skip to Navigation
Conference papers

Probabilistic Model of Error in Fixed-Point Arithmetic Gaussian Pyramid

Antoine Meler 1 John Alexander Ruiz Hernandez 1 James L. Crowley 1
1 PRIMA - Perception, recognition and integration for observation of activity
Inria Grenoble - Rhône-Alpes, UJF - Université Joseph Fourier - Grenoble 1, INPG - Institut National Polytechnique de Grenoble , CNRS - Centre National de la Recherche Scientifique : UMR5217
Abstract : The half-octave Gaussian pyramid is an important tool in computer vision and image processing. The existence of a fast algorithm with linear computational complexity makes it feasible to implement the half-octave Gaussian pyramid in embedded computing systems using only integer arithmetic. However, the use of repeated convolutions using integer coefficients imposes limits on the minimum number of bits that must be used for representing image data. Failure to respect this limits results in serious degradation of the signal to noise ratio of pyramid images. In this paper we present a theoretical analysis of the accumulated error due to repeated integer coefficient convolutions with the binomial kernel. We show that the error can be seen as a random variable and we deduce a probabilistic model that describes it. Experimental and theoretical results demonstrate that the linear complexity algorithm using integer coefficients can be made suitable for video rate computation of a half-octave pyramid on embedded image acquisition devices.
Document type :
Conference papers
Complete list of metadatas
Contributor : Patrick Reignier <>
Submitted on : Friday, February 28, 2014 - 11:54:40 AM
Last modification on : Wednesday, April 11, 2018 - 1:53:48 AM

Links full text




Antoine Meler, John Alexander Ruiz Hernandez, James L. Crowley. Probabilistic Model of Error in Fixed-Point Arithmetic Gaussian Pyramid. 5th IEEE Workshop on Embedded Computer Vision (held in conjuction with ICCV 2009), 2009, Kyoto, Japan. pp.916-820, ⟨10.1109/ICCVW.2009.5457618⟩. ⟨hal-00953484⟩



Record views