Gavin S. Page


The human's innate ability to process information garnered from a visual scene has no parallel in the digital realm. This task is taken for granted in human cognition, but has not been met by a complete digital solution even following years of research. This difficulty can be explained by the shear complexity of the physology of the visual pathway. Although a complete solution has not been created, there are a number of examples of solutions that address parts of the problem. The recognition of deformable objects is the area addressed in this work. The specific task researched was the recognition of creatures in structured visual scenes. The focus was on developing a set of features which are able to differentiate between target creature classes. The implications of this research lie in ecoinformatics and field biology with the automated collection and annotation of biological data. The thesis will present a survey of the current literature addressing techniques which have been used to solve similar problems. An algorithm to perform the recognition will be presented and the results discussed. Finally, potential areas for improvement will be described.

Library of Congress Subject Headings

Computer vision; Pattern recognition systems; Deformation potential; Image processing--Digital techniques

Publication Date


Document Type


Student Type


Degree Name

Computer Science (MS)

Department, Program, or Center

Computer Science (GCCIS)


Gaborski, Roger


Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TA1634 .P34 2005


RIT – Main Campus