Image Analysis of Soil Micromorphology: Feature Extraction, Segmentation, and Quality Inference
EURASIP Journal on Applied Signal Processing
Image Analysis of Soil Micromorphology: Feature Extraction, Segmentation, and Quality Inference
Petros Maragos 0 1
Giorgos B. Stamou 0 1
Vassilis Tzouvaras 0 1
Efimia Papatheodorou
0 School of Electrical & Computer Engineering, National Technical University of Athens , Athens 15773 , Greece
1 Department of Biology, Ecology Division, Aristotle University of Thessaloniki , Thessaloniki 54006 , Greece
We present an automated system that we have developed for estimation of the bioecological quality of soils using various image analysis methodologies. Its goal is to analyze soilsection images, extract features related to their micromorphology, and relate the visual features to various degrees of soil fertility inferred from biochemical characteristics of the soil. The image methodologies used range from low-level image processing tasks, such as nonlinear enhancement, multiscale analysis, geometric feature detection, and size distributions, to object-oriented analysis, such as segmentation, region texture, and shape analysis.
and phrases; soilsection image analysis; geometric feature extraction; morphological segmentation; multiscale texture analysis; neurofuzzy quality inference
1. INTRODUCTION
The goal of this research work is the automated estimation
of the bioecological quality of soils using image processing
and computer vision techniques. Estimating the soil quality
with the traditional biochemical methods, and more
specifically estimating those elements that are essential
components for the soil fertility, is a difficult, time-consuming, and
expensive process, which is, however, necessary for
selecting and applying any management practice to land
ecosystems. Our approach has been the development of an
automated system that will recognize the characteristics relevant
to the soil quality by computer processing of soilsection
images and extraction of suitable visual features. Its final goals
are double-fold: (
1
) quantification of the micromorphology
of the soil via analysis of soilsection images and (
2
)
correspondence of the extracted visual information with the
classification of soil into various fertility degrees inferred from
measurements performed biochemically on the soil samples.
The overall system is shown in Figure 1.
Soil
(map image)
Soil sampling
Digital image
acquisition system
(digital camera,
scanner)
Filtering for image
enhancement
Watershed
segmentation
Multiscale image
analysis
Geometrical
feature
extraction
Marker
detection/
extraction
Texture analysis
with AM-FM
models
Size distribution
histograms
and moments
measures
Shape analysis
Homogeneous
regions texture
analysis with
fractals
Chemical analysis
Neural network
Initial knowledge
of soil quality
from soilsection
features
Feature
extraction
using computer
vision
Correspondence
Soil quality
evaluation
In the image analysis part of this work, the above goals
require solving a broad spectrum of problems in image
processing and computer vision. Next, we list the most
important of such problems (following a hierarchy from
low-level vision to high-level vision) which we have
investigated for detecting characteristics and extracting
information from soilsection images: (
1
) enhancement of
images; (
2
) feature detection; (
3
) multiscale image analysis;
(
4
) statistical size distributions; (
5
) segmentation into
homogeneous regions; (
6
) texture analysis; (
7
) shape
analysis; and (
8
) correspondence of the features extracted from
analyzing the soilsection images with the fertility grade of
the soil inferred from its biochemical characteristics. The
tools and methodologies that we have used for solving the
above image analysis problems (
1
)–(
7
) include the
following: (i) nonlinear geometric multiscale lattice-based image
operators (of the morphological and fuzzy type) for
multiscale image simplification and enhancement, extracting
presegmentation features, size distributions, and region-based
segmentation; (ii) nonlinear partial differential equations
(PDEs) for isotropic modeling and implementing various
multiscale evolution and visual detection tasks; (iii)
fractals for quantifying texture and shape analysis from the
viewpoint of geometrical complexity; (iv) modulation
models for texture modeling from the viewpoint of
instantaneous spatial frequency and amplitude components; and
(v) topological and curvature-based methods for region
shape analysis. Finally, methods of fuzzy logic and
neural networks were investigated for the symbolic
description and automated adaptation of the correspondence
between the soilsection images and the bioecological quality of
soil.
2. SOIL DATA AND MICROMORPHOLOGY
Soil data: the first phase of this work dealt with collecting
soil samples both for performing biochemical measurements
and for computer-based automated analysis of their images.
During the phase of data collection, soil was sampled in
mid September 2000 und (...truncated)
