Tuning a complete image processing chain (IPC) is not a straightforward task. The first problem to overcome is the evaluation of the whole process. Until now researchers have focused on the evaluation of single algorithms based on a small number of test images and ad hoc tuning independent of input data. In this paper, we explain how the design of experiments applied on a large image database enables statistical modeling for IPC significant parameter identification. The second problem is then considered: how can we find the relevant tuning and continuously adapt image processing to input data? After the tuning of the IPC on a typical subset of the image database using numerical optimization, we develop an adaptive IPC based on a neural network working on input image descriptors. By testing this approach on an IPC dedicated-to-road obstacle detection, we demonstrate that this experimental methodology and software architecture can ensure continuous efficiency. The reason is simple: the IPC is globally optimized, from a large number of real images and with adaptive processing of input data.
Description
Design of experiments for performance evaluation and parameter tuning of a road image processing chain
%0 Journal Article
%1 Lucas:2006:DEP:1288263.1288455
%A Lucas, Yves
%A Domingues, Antonio
%A Driouchi, Driss
%A Treuillet, Sylvie
%C New York, NY, United States
%D 2006
%I Hindawi Publishing Corp.
%J EURASIP J. Appl. Signal Process.
%K ComputerVision algorithms benchmark evaluation performance statistics to_READ
%P 212--212
%R 10.1155/ASP/2006/48012
%T Design of experiments for performance evaluation and parameter tuning of a road image processing chain
%U http://dx.doi.org/10.1155/ASP/2006/48012
%V 2006
%X Tuning a complete image processing chain (IPC) is not a straightforward task. The first problem to overcome is the evaluation of the whole process. Until now researchers have focused on the evaluation of single algorithms based on a small number of test images and ad hoc tuning independent of input data. In this paper, we explain how the design of experiments applied on a large image database enables statistical modeling for IPC significant parameter identification. The second problem is then considered: how can we find the relevant tuning and continuously adapt image processing to input data? After the tuning of the IPC on a typical subset of the image database using numerical optimization, we develop an adaptive IPC based on a neural network working on input image descriptors. By testing this approach on an IPC dedicated-to-road obstacle detection, we demonstrate that this experimental methodology and software architecture can ensure continuous efficiency. The reason is simple: the IPC is globally optimized, from a large number of real images and with adaptive processing of input data.
@article{Lucas:2006:DEP:1288263.1288455,
abstract = {Tuning a complete image processing chain (IPC) is not a straightforward task. The first problem to overcome is the evaluation of the whole process. Until now researchers have focused on the evaluation of single algorithms based on a small number of test images and ad hoc tuning independent of input data. In this paper, we explain how the design of experiments applied on a large image database enables statistical modeling for IPC significant parameter identification. The second problem is then considered: how can we find the relevant tuning and continuously adapt image processing to input data? After the tuning of the IPC on a typical subset of the image database using numerical optimization, we develop an adaptive IPC based on a neural network working on input image descriptors. By testing this approach on an IPC dedicated-to-road obstacle detection, we demonstrate that this experimental methodology and software architecture can ensure continuous efficiency. The reason is simple: the IPC is globally optimized, from a large number of real images and with adaptive processing of input data.},
acmid = {1288455},
added-at = {2013-10-22T15:59:03.000+0200},
address = {New York, NY, United States},
author = {Lucas, Yves and Domingues, Antonio and Driouchi, Driss and Treuillet, Sylvie},
biburl = {https://www.bibsonomy.org/bibtex/2d6f469541e30dfe068f1909130a7c3a0/alex_ruff},
description = {Design of experiments for performance evaluation and parameter tuning of a road image processing chain},
doi = {10.1155/ASP/2006/48012},
interhash = {ec531ec7c01ea3633c16f589abf6effb},
intrahash = {d6f469541e30dfe068f1909130a7c3a0},
issn = {1110-8657},
issue_date = {01 January},
journal = {EURASIP J. Appl. Signal Process.},
keywords = {ComputerVision algorithms benchmark evaluation performance statistics to_READ},
month = jan,
numpages = {1},
pages = {212--212},
publisher = {Hindawi Publishing Corp.},
timestamp = {2013-10-22T15:59:03.000+0200},
title = {Design of experiments for performance evaluation and parameter tuning of a road image processing chain},
url = {http://dx.doi.org/10.1155/ASP/2006/48012},
volume = 2006,
year = 2006
}