Near sets in pattern similarity distance based classification
No Thumbnail Available
This research is focused on studying the nearness theory, various neighbourhoods of points in proximity spaces and similarity measures which leads to discovering the set patterns in digital images. The problem considered in this thesis is set pattern discovery using topology of digital images and nearness theory and then using the extracted set patterns as the basis for the classification method that is introduced. Visual pattern is the repetition of some forms in a digital image. A visual set pattern is defined as a collection of sets that all of the members of the collection have common features and properties and are all descriptively near a given set called the pattern generator or motif set which itself is a bounded descriptive neighbourhood of a distinguished point of interest in the image. Using the generated set patterns, a classification method is introduced based on the set patterns similarity distance.
Nearness theory, Neighbourhoods, Proximity space, feature vector, Probe function, Spatial, Descriptive, Set pattern, Set pattern generation, Motif, Pattern, Pattern recognition, Similarity Measure, Pattern similarity distance, Saliency, Classification