Application of stochastic geometry methods to pattern recognition is analyzed. The paper is based on Trace-transforms of original images introduced by [1] into images on the Möbius band. The ability of a Trace-transform to solve such structuralistic problems as segmentation, analysis of objects' relative position, and their number evaluation, is established. Feasibility of image nonlinear filtering through Trace-transforms is considered. Based on the new geometric transform, a new approach towards the construction of features, independent of images' motions or their linear transforms, is put forward. A prominent characteristic of the group of features under consideration is that we can represent each of them as a consecutive composition of three functionals. © Springer-Verlag 2003.
CITATION STYLE
Fedotov, N., & Shulga, L. (2003). New geometric transform based on stochastic geometry in the context of pattern recognition. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2749, 148–155. https://doi.org/10.1007/3-540-45103-x_21
Mendeley helps you to discover research relevant for your work.