Authors:
Cindy Torres
;
Alain Clément
and
Bertrand Vigouroux
Affiliation:
Institut Universitaire de Technologie, France
Keyword(s):
Color, Spatial Organization, Classification, Segmentation, Plants.
Related
Ontology
Subjects/Areas/Topics:
Computer Vision, Visualization and Computer Graphics
;
Image and Video Analysis
;
Segmentation and Grouping
Abstract:
An unsupervised vectorial segmentation method using both spatial and color information is presented. To overcome the problem of memory space, this method is based on a multidimensional compact histogram and an original compact spatial neighborhood probability matrix (SNPM). The multidimensional compact histogram allows a drastic reduction of memory space without any data loss. Leaning upon the compact histogram, a SNPM has been computed. It contains all non-negative probabilities of spatial connectivity between pixel colors. In an unsupervised histogram analysis classification process, two phases are classically distinguished: (i) a learning process during which histogram modes are identified and (ii) a second step called the decision step in which a full partition of the colorimetric space is carried out according the previously defined classes. During the second step of a standard colorimetric approach, a colorimetric distance like Euclidean or Mahalanobis is used. We insert here a
spatio-colorimetric distance defined as a weighed mixture between a colorimetric distance and the spatial distance calculated from the SNPM. The vectorial classification method is based on previously presented principles, achieving a hierarchical analysis of the color histogram by means of a 3D-connected components labeling. Results are applied to color images of plants to separate plantlets and loam.
(More)