- Accueil
- Volume 12 (1993)
- Number 2 - Proceedings of the sixth European congr...
- Grey level image analysis using a topological tree of connected figures selected in a set of serial binary sections
Visualisation(s): 320 (6 ULiège)
Téléchargement(s): 36 (3 ULiège)
Grey level image analysis using a topological tree of connected figures selected in a set of serial binary sections
Abstract
A method of grey level image analysis is proposed. It consists in the extraction and labeling of connected components (figures) in a set of serial binary sections, establishing connectivity between figures in adjacent sections, forming of image description, building of a topological tree of connected figures. The topological tree is analyzed for extraction of objects of interest. The parameters which characterize the location and size are calculated for each figure. Topological tree analysis can be made taking into account the figure dimensional parameters and the tree furcations, which characterize grey level image topological features. Parameters filtering allows to erase the information about those figures which do not satisfy the appropriate constraints, including the figures which extraction is the result of noise. For extracted objects of interest the extended set of parameters (area, perimeter, coordinates of gravity center, shape and orientation parameters and so on) is calculated, which may be used for image recognition.
The method was realized in the image processing system consisting of a light microscope, a TV camera, a Frame Grabber card built in IBM PC 386/ 20 Mhz and image processing software package. The system enables to fulfil segmentation of a grey level image by 512 x 512 x 8 format using 64 binary sections for the time which does not exceed 35 seconds. This method is effective for the extraction of low contrast objects located at uneven background. The method was used for image analysis of glial cells in the substantia nigra of infant brains.