Ekstraksi Fitur Menggunakan Elliptical Fourier Descriptor Untuk Pengenalan Varietas Tanaman Kedelai

摘要: Hermawan Syahputra. Feature Extraction Using Elliptic Fourier Descriptor for Recognition Soybean Varieties. Under direction of AGUS BUONO, YENI HERDIYENI and SUTORO. One real problem in agriculture or biology is difficulty to identify soybean (Glycine max (L.) merrill) varieties. (EF) analyses were used varieties, based on leaf shape. Chain encoded, harmonic functions generated boundary. Leaves varieties was evaluated quantitatively by image analysis using elliptic descriptors. The closed contour each projection extracted, coefficients calculated contour. standardized so that they invariant size, rotation, shift, chain code starting point. classification results showed identified with an average correct rate 40%. This result achieved combination top-centre leaf, top-beside middle-beside leaf. reason why the accuration low are leaves have high variation one variety. shape feature could be important accurate tool identification mapping. Keyword : Image processing, Descriptor, Chaincode, extraction, pattern recognition.