BLIND SEPARATION IN LOW FREQUENCIES USING WAVELET
ANALYSIS, APPLICATION TO ARTIFICIAL VISION
Danielle Nuzillard*, Sorin Curila*
,
**, Mircea Curila*
,
**
Laboratoire d'Automatique et de Microélectronique*, Université de Reims Champagne-Ardenne, Campus du Moulin de la Housse
B.P. 1039, 51687 REIMS Cedex 2, FRANCE, Tel: +33.(0)3.26.91.82.16, Fax : +33.(0)3.26.91.31.06
University of Oradea**, 3700 Oradea, ROMANIA, str. Armatei Romane 5
e-mail: danielle.nuzillard@univ-reims.fr, sorin.curila@univ-reims.fr, mircea.curila@univ-reims.fr
ABSTRACT
We propose a method for image enhancement in colour word
when a scattering environment reduces the vision. The main
advantage of blind technique is that it does not require any a
priori information about the scattering environment but supposes
that the observed signals are linear mixtures of sources. Here, the
natural logarithm of the degraded image provides an
approximative additive mixture of reflectivity and transmittivity
coefficients, the colour images provide three coloured mixtures
(red, green, blue). They are processed by a Blind Source
Separation (BSS) method in low spatial frequencies to display
gray levels of pertinent features, which help one to vision
enhancement. To display a cleaner vision, the set of mixtures is
enriched thanks to classical signal processing technique. The
chrominance information is restituted using post-processing
techniques on HSV (Hue, Saturation, Value) space of degraded
colour image. Experiments are made on images for which
scattering environment is simulated in the laboratory.
Keywords- Scattering environment, Artificial vision, Blind
Source Separation, Second order blind identification, Independent
component analysis, Wavelet denoising
1. INTRODUCTION
The goal of computer vision is to develop techniques
applied to artificial vision for which performance matches
or exceeds that of human. Computer vision systems must
include methods that enable applications for detecting
objects through scattering environment, in bad weather
conditions such as fog, rain, haze, snow, etc. The proposed
topic is scientifically innovative because it associates the
classical physics (radiative transfer in the scattering
environment) to modern signal and image processing
techniques as Blind Source Separation (BSS). The
application of BSS methods for the improving the vision in
fog conditions has for justification that the statistical
properties of the scene are different from those of the
environment between the scene and the observer.
In the literature, the atmospheric propagation and the
distribution of particles participating in effects such as light
scattering correspond to an atmospheric model. We chosed
to use a simple description that yields the properties of an
actually model but only takes care of the image contrast
degradation by surface radiation and atmospheric radiation
phenomena.
The simple model is described in section 2. To
enhance the vision, we propose an approach presented in
sections 3. The searched sources, using BSS methods, are
the depth distribution (related to the transmittivity
coefficient for an homogeneous scattering environment)
and the reflectivity coefficient of the scene. The separation
takes place in low frequencies domain due to environment
distribution. Histogram equalization and its variations are
very flexible since no assuming is made about the physics
of the scene. But the results of these methods suggest that
they are less effective in enhancing degraded images under
scattering environment than model-based methods. The
performances of the proposed method show a gain of 25%
compared to the histogram equalization. Image processing
techniques on HSV (Hue, Saturation, Value) space,
proposed in section 4, complete the enhanced process to
yield a colour image. Experimental results are available in
section 5.
An intense research activity falls out to describe the
scattering environment properties and to enhance the image
quality under such conditions. We mention here those
whose aims are very close to our work.
National Oceanic and Atmospheric Administration
(NOAA) Science Center [19] catches images from two
geostationary operational environmental satellites and
combines them to create a product for the detection of fog
and low status clouds at night. The "Fog Image" is
obtained by a subtraction of the short-wave infrared
window channel (3.9 micron wavelength) temperatures
from the long-wave infrared window channel (10.7
micron) temperatures. The image is a special enhancement
that estimates the thickness of a low cloud layer, based on
the temperature difference between the two infrared
channels.
S. K. Nayar studies the visual manifestations of
different weather conditions. Based on his observations, he
develops models and methods for recovering pertinent
scene properties such as three-dimensional structure from
images taken under poor weather conditions [13].
D. Jackel presents a method for rendering of the
atmosphere by means of Mie-scattering. The method is
77
4th International Symposium on Independent Component Analysis and Blind Signal Separation (ICA2003), April 2003, Nara, Japan