Date and time: 11.12.2018 14:15-15:00
Speaker: Slobodan Dražić, University of Novi Sad, Serbia
Title: Some digital image processing methods for shape-based object comparison
Slides can be found here (TBA)
Methods for quantitative characterization of objects from their digital images are increasingly used in applications in which an error can have critical consequences, such as medical imaging or autonomous vehicle control. Still, the traditional methods for shape quantification are of low precision and accuracy. The traditional methods are based on binary images in which pixels are crisply divided on those that completely belong to the interior or border of an object and those that completely belong to the exterior. We show that the additional information about coverage of every pixel by an object can be successfully used to improve usage of digital images to compute the maximal distance between objects furthest points measured in a given direction.
It is highly desirable that a distance measure between digital images can be related to a certain shape property. Morphological operations are used when defining a distance for this purpose. Still, the distances defined in this manner turns out to be insufficiently sensitive to the relevant data representing shape properties in images. We show that the idea of adaptive mathematical morphology can be successfully used to overcome problems related to the sensitivity of distances defined via morphological operations when comparing objects from their digital image representations.