Markov Random Field and Fuzzy Logic Modeling in Sonar Imagery: Application to the Classification of Underwater Floor

doi:10.1006/cviu.2000.0844

Computer Vision and Image Understanding

Volume 79, Issue 1, July 2000, Pages 4-24

https://doi.org/10.1006/cviu.2000.0844 Get rights and content

Abstract

This paper proposes an original method for the classification of seafloors from high resolution sidescan sonar images. We aim at classifying the sonar images into five kinds of regions: sand, pebbles, rocks, ripples, and dunes. The proposed method adopts a pattern recognition approach based on the extraction and the analysis of the cast shadows exhibited by each seabottom type. This method consists of three stages of processing. First, the original image is segmented into two kinds of regions: shadow (corresponding to a lack of acoustic reverberation behind each “object” lying on the seabed) and seabottom reverberation. Second, based on the extracted shadows, shape parameter vectors are computed on subimages and classified with a fuzzy classifier. This preliminary classification is finally refined thanks to a Markov random field model which allows to incorporate spatial homogeneity properties one would expect for the final classification map. Experiments on a variety of real high-resolution sonar images are reported.

References (29)

B. Chalmond
An interative Gibbsian technique for reconstruction of M-ary images
Pattern Recognition
(1989)
F. Salzenstein et al.
Parameter estimation in hidden fuzzy Markov random fields and image segmentation
Graph. Models Image Process.
(1997)
D. Alexandrou and D. Pantzartzis, Seafloor classifcation with neural networks, in Proc. OCEANS, Washington, D.C., 1990,...
D. Alexandrou et al.
A methodology for acoustic seafloor classification
IEEE J. Oceanic Eng.
(1993)
S. Banks
Signal Processing Image Processing and Pattern Recognition,
(1990)
J. Besag
On the statistical analysis of dirty pictures
J. R. Stat. Soc. B
(1986)
B. Bourgeois and C. Walker, Sidescan sonar image interpretation with neural networks, in Proc. OCEANS, Honolulu, 1991,...
D. Carmichael et al.
Seabed classification through multifractal analysis of sidescan sonar imagery
IEE Proc. Radar Sonar Nav.
(1996)
D. Cobra and H. Moraes, Classification of sidescan sonar images through parametric modeling, in Proc. OCEANS, Brest,...
C. Collet, J.-M. Burel, and E. Borderie, Multiscale discriminant analysis for texture classification on high resolution...

H. Derin et al.

Modeling and segmentation of noisy and textured images using Gibbs random fields

IEEE Trans. Pattern Anal. Mach. Intell.

(1987)

F. Forbes, and, A. E. Raferty, Bayesian morphology: Fast unsupervised Bayesian image analysis, J. Am. Stat. Assoc, to...

S. Geman et al.

Stochastic relaxation, Gibbs distributions and the Bayesian restoration of images

IEEE Trans. Pattern Anal. Mach. Intell.

(1984)

J. Goodman

Some fundamental properties of speckle

J. Opt. Soc. Am.

(1976)

Cited by (34)

The measure on the original space from a product measure
2020, Journal of Mathematical Analysis and Applications
Citation Excerpt :
A few major specific examples of image processing using MRF includes image segmentation and skin detection as part of a system for human emotion recognition from color images [13], image processing of multi-dimensional multivariate Gaussian Markov random fields [14], segmentation of images relying on Bayesian smoothing [5], image retrieval task involving global image features [12], segmentation of brain MR images [11], image segmentation via self-validated labeling of Markov random fields [6]. MRFs were also applied to such exciting areas as sonar signal analysis for the classification of underwater floor [15], landmine detection in ground penetrating radar data [19], fast damage mapping in case of earthquakes [20], and automatic detection of pornographic images [25]. Furthermore, as special cases of MRFs, Hidden Markov Models (HMMs) and dynamic Bayesian Networks (BNs) were intensively applied to such areas as speech recognition (e.g., [9]), natural language processing (e.g., [1]), handwriting recognition (e.g., [26]), and bioinformatics (e.g., [2]).
The measurability problems can arise in Gaussian and empirical processes underlying nonparametric and semiparametric problems critical in modern statistics and machine learning. The usual applications are high-dimensional data in random fields. To ensure the validity of recovering a true original complex object from its observed images treated in a product space that appears in existing high-dimensional data applications widely arising in physical, biological, and social sciences and engineering, we begin with the results on the σ-algebra of the original space Ω induced by a given map $f_{T}$ mapping from Ω to a product measure space $(Ω_{T}, F_{T}, μ_{T})$ . We study the certain measurability issues on Ω with the induced measure μ from the image measure $μ_{T}$ of the product space by establishing the outer measure $ν^{⁎}$ restricted to the trace σ-algebra. We further investigate the measure μ on Ω under the map $f_{T}$ to obtain the conditions for the identifiability equality $μ_{T} = μ \circ f_{T}^{- 1}$ to hold so that a practical and rigorous foundation is provided for the nontrivial applications to work meaningfully.
Lightweight YOLOv5 sonar image object detection algorithm and implementation based on ZYNQ
2024, Guangdian Gongcheng/Opto-Electronic Engineering
Saliency Map Estimation Using a Pixel-Pairwise-Based Unsupervised Markov Random Field Model
2023, Mathematics
Benthic Organism Detection, Quantification and Seamount Biology Detection Based on Deep Learning
2023, Artificial Intelligence Oceanography
MEASURE-THEORETIC PROBABILITY OF COMPLEX CO-OCCURRENCE AND E-INTEGRAL
2022, arXiv
Dual-Path Residual "shrinkage" Network for Side-Scan Sonar Image Classification
2022, Computational Intelligence and Neuroscience

View all citing articles on Scopus

^☆: The authors thank GESMA (“Groupe d'Étude Sous-Marine de l'Atlantique,” Brest, France), for having provided numerous real sonar pictures, and DGA (“Direction Générale de l'Armement,” French Ministry of Defense) for financial support of this work via student grant.

^f1: [email protected]

View full text

Regular ArticleMarkov Random Field and Fuzzy Logic Modeling in Sonar Imagery: Application to the Classification of Underwater Floor☆

Abstract

Pattern Recognition

Graph. Models Image Process.

A methodology for acoustic seafloor classification

IEEE J. Oceanic Eng.

Signal Processing Image Processing and Pattern Recognition,

On the statistical analysis of dirty pictures

J. R. Stat. Soc. B

Seabed classification through multifractal analysis of sidescan sonar imagery

IEE Proc. Radar Sonar Nav.

Modeling and segmentation of noisy and textured images using Gibbs random fields

IEEE Trans. Pattern Anal. Mach. Intell.

Stochastic relaxation, Gibbs distributions and the Bayesian restoration of images

IEEE Trans. Pattern Anal. Mach. Intell.

Some fundamental properties of speckle

J. Opt. Soc. Am.

Regular Article
Markov Random Field and Fuzzy Logic Modeling in Sonar Imagery: Application to the Classification of Underwater Floor☆