期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2021
卷号：99
期号：18
语种：English
出版社：Journal of Theoretical and Applied
摘要：Biodiversity is the strength of an ecosystem. Indonesia is a country that has a diversity of ornamental fish species both marine and freshwater ornamental fish. Maintaining its sustainability by identifying its organisms is important. Identification is important because all subsequent work sequences depend on the correct identification results. Identification is looking for and recognizing individual taxonomic characteristics and incorporating them into a taxon. The high diversity becomes its own obstacle in conducting a fast and precise identification. The computer aided of fish�s identification and classification by using image processing technology has been widely carried out. The results of previous studies indicate the identification and classification of the fishes are narrowed only to the stage of distinguishing the characteristics of fish species at the family level. This research tries to identify and classify more deeply to be able to distinguish 3 species in one genus Melanotaenia (Rainbow fish).This study extracts features based on morphometric network truss using distance measurement methods and produces features in the form of the ratio of segment length ratio toward standard length of the Rainbow fish. The learning method used for classification and introduction is the Linear Discriminant Analysis method.The results showed that the distance measurement method can be used to obtain the characteristics of the Rainbow fish based on morphometric network truss, which is independent toward the fish�s size and the magnification of objects in the image. Another result is that the Linear Discriminant Analysis method can produce an accuracy of 70.83% to 75.00% in the classification of Rainbow fish species.
关键词：Feature Extraction;Morphometric Truss;Linear Discriminant Analysis;Rainbow Fi
