A string matching computer-assisted system for dolphin photoidentification.

摘要: This paper presents a syntactic/semantic string representation scheme as well matching method part of computer-assisted system to identify dolphins from photographs their dorsal fins. A low-level is constructed the curvature function dolphin's fin trailing edge, consisting positive and negative primitives. high-level then built over via merging appropriate groupings primitives in order have less sensitive fluctuations or noise. family distance measures between two strings introduced. composite measure defined used dissimilarity for database search, highlighting both syntax (structure sequence) semantic (attribute feature) differences. The consists an ordered sequence significant protrusions intrusions on while semantics consist seven attributes extracted edge its function. results are reported 624 images corresponding 164 individual dolphins. identification indicate that developed performs better than previous methods including ratio, curvature, curve matching. can help marine mammalogists dolphins, since it allows them examine only handful candidate instead currently manual searching entire database. © 2000 Biomedical Engineering Society. PAC00: 8780Tq, 4230Sy, 0705Pj