Teknologi Pengolahan Citra Digital Untuk Ekstraksi Ciri pada Citra Daun untuk Identifikasi Tumbuhan Obat

Abstract

The leaf image identification process depends on the feature extraction results. Each medicinal plant has different shapes and patterns of leaf venation. But for one type of medicinal plant have the same pattern of venation shape and pattern even though the size is different. One of the methods for extraction of leaf image form characteristics is fractal-based feature extraction. Through fractal can be calculated the value of leaf dimensions and searched parts of leaves that have similarities between one part with other parts. As for the method of extracting the characteristics of venation pattern using the B-Spline method. The benefits of research conducted are to help people identifying the types of medicinal plants found, knowing the benefits and ways of brewing. While the research contribution is a prototype software application based on information technology that can be used by people through mobile phones for the identification of medicinal plants. To identify or match the results of feature extraction on the leaf found whether included in the medicinal plant, conducted by Euclidean Distance method. In the experiments we used 1100 data consist of 55 varieties of medicinal plants for every 20 samples. The experimental result shows that the accuracy of identification using fractal and b-spline is 85.30%.