METHODS FOR ACQUIRING SHAPES FROM HEP-2 CELL SECTIONS AND THE CASE-BASED RECOGNITION OF HEP-2 CELLS

摘要

The invention relates to methods for acquiring shapes from images with representations of HEp-2 cell sections in the form of objects and for learning abstract shape models from representations of HEp-2 cell sections for a case database for a case-based recognition of HEp-2 cells in digital images. The invention also relates to methods for acquiring shapes from images with representations of HEp-2 cell sections in the form of cases and for the case-based recognition of HEp-2 cells in the form of objects in digital images, to computer program products having a program code for carrying out these methods, to computer program products on machine-readable carriers for carrying out these methods, and to digital storage media that can interact with a programmable computer system whereby carrying out these methods. The methods are characterized in that individual shapes of HEp-2 cell sections are semiautomatically collected as objects in the form of representations in images and in that abstract shape models in different abstraction levels can be automatically obtained from these individual shapes. The learned abstract shape models are either averaged shapes from groups of groups of objects or medians in the form of individual groups of HEp-2 cells.