EDGE DETECTION FOR BUTTERFLY SPECIES CLASSIFICATION USING SUPPORT VECTOR MACHINE (SVM)

Nor Hazirah, Mohd Tamin (2019) EDGE DETECTION FOR BUTTERFLY SPECIES CLASSIFICATION USING SUPPORT VECTOR MACHINE (SVM). [Final Year Project Report] (Unpublished)

PDF (Please get the password from TECHNICAL & DIGITIZATION MANAGEMENT UNIT, ext: 082-583913/ 082-583914) Nor Hazirah Binti Mohd Tamin.pdf Restricted to Registered users only Download (2MB)

Abstract

This research project is to help taxonomists in classifying species of butterfly with a modern method. Previously, taxonomists used traditional ways to identify the species, but it has several drawbacks such as the specimen being easily damaged over time, pests destroying the collection or discolouration of the specimen. Thus, with modern technology which in this project used a support vector machine as for classifying the species, help in solving the drawbacks. Before classifying the butterfly species, edge detection for the wing structure of butterflies is performed. Edge detection is one of image processing techniques, helping in showing clearly the recognition features of each species. The butterfly species can be recognized by using their recognition features. The recognition features of the butterflies is taken as parameter in determining their species. Recognition features involved in this project are mean, standard deviation and entropy of images. Support vector machine is used for classification purposes and a well- trained SVM can classify the species based on the recognition features extracted from edge detection. For final year project part 1, the images of both butterfly species are collected and used for performing the edge detection process. The recognition features for both species is determined as parameters for the classifying process using a support vector machine. While for final year project part 2, the process involved in extraction of the recognition features before those data undergoes training process and test process. The training process is to allow SVM to learn about the data and build a model. The model is used with the test set in order to classify the species.

Actions (For repository members only: login required)

View Item