This project will construct a database of dorsal fin images of Tursiops truncatus (bottlenose dolphins) sighted near the Texas and California coasts, and software to automatically identify individual dolphins based on fin morphology. The system will be used to match individuals in new photographs to existing known individuals by finding the database member that most resembles the new input data. An existing resource of many thousands of images of dolphins will be processed by the system and will constitute the largest searchable database of dolphin sightings available. The database will provide new information regarding the site fidelity and migratory patterns of dolphins. The system will be developed for immediate use in marine biological research, and will be demonstrated and tested using images of the dorsal fins of dolphins. However, the system will be sufficiently versatile for many other applications requiring identification of individual animals. The project will be conducted by two collaborating teams of investigators: computer scientists and biologists. Each team has specific objectives, but their coordinated work will result in a computer-based biological research tool that will be useful to investigators conducting studies requiring the identification of individuals within a population. The computational objectives of the project will be to: 1.Extract the individuals' fin shapes from photographic data; 2.Calculate of characteristics of the shape that can be used to distinguish one individual from another; 3.Create of a database of images, indexed by the features extracted from the images; 4.Provide capabilities for entering data, building, searching, and modifying databases, with a convenient user interface. The software system will be offered to biologists needing to identify individuals of any species on the basis of photographic data. The image processing portion of the system will be suitable for recognizing a profile of a bo dy part, such as a fin or an ear, having distinctive irregularities. The database system, however, will not be limited to descriptions of profile shapes, but will be applicable to any situation where some distinctive aspect of an animals appearance can be measured. The dolphin database will be made available to investigators interested in behavioral biology of cetaceans, and to ecologists interested in the environmental influences acting upon abundance and distribution patterns.
