Premium
APPLICATION OF THE GABOR FILTERS AND K ‐MEANS METHOD FOR CHARACTERIZATION OF GEOMETRICAL PROPERTIES OF PARENCHYMA CELL WALLS
Author(s) -
KURITA H.,
MASUDA R.
Publication year - 2012
Publication title -
journal of texture studies
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.593
H-Index - 54
eISSN - 1745-4603
pISSN - 0022-4901
DOI - 10.1111/j.1745-4603.2012.00345.x
Subject(s) - gabor filter , texture (cosmology) , orientation (vector space) , filter (signal processing) , segmentation , mesoscopic physics , artificial intelligence , characterization (materials science) , biological system , pattern recognition (psychology) , shape factor , mathematics , computer science , geometry , computer vision , feature extraction , image (mathematics) , optics , physics , quantum mechanics , biology
It is important to analyze the geometrical features of cell wall arrangements to gain further insight into the mechanical properties of plant parenchyma. To extract such features, we conduct a Gabor filter analysis. A two‐dimensional Gabor filter can detect local textual features because of its selective sensitivity to stimuli of specified directionality and interval. This study first discusses the extraction of cell wall arrangement from cross‐sectional images of carrot parenchyma using a Gabor filter. Second, using the k ‐means method, we investigate the distribution of those cell walls found to have a particular orientation and interval in the first analysis. We propose a new method to extract the mesoscopic geometrical properties of cell walls such as orientation and interval, distribution and, in some cases, shape. This method is applied to some exemplar images, and the result is demonstrated in images and spectral graphs. PRACTICAL APPLICATIONS It is important to research the relation between the mechanical and geometrical properties of cell walls in terms of the appropriate handling and storage of crops, assessing crop quality, proper soil conditioning for root crops and improving food texture or forage. The method presented in this article is useful for extracting geometrical properties from irregularly textured images and will be helpful in studying the relation between the two types of properties.