MORPHOLOGICAL AND TEXTURAL FEATURE EXTRACTIONS FROM FUNGI IMAGES FOR DEVELOPMENT OF AUTOMATED MORPHOLOGY-BASED FUNGI IDENTIFICATION SYSTEM

R. Putri Ayu Pramesti (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Muhamad Rodhi Supriyadi (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Aulia Haritsuddin Karisma (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Muhammad Reza Alfin (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Mukti Wibowo (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Bayu Rizky Maulana (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Gilang Mantara Putra (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Josua Geovani Pinem (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Umi Chasanah (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Kristiningrum Kristiningrum (Research Center for Vaccine and Drug, BRIN)
Ariza Yandwiputra Besari (Research Center for Vaccine and Drug, BRIN)
Avi Nurul Oktaviani (Research Center for Applied Microbiology (PRMikTer), BRIN)
Dyah Noor Hidayati (Research Center for Applied Microbiology (PRMikTer), BRIN)
Dewi H Budiarti (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Jemie Muliadi (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)
Danang Waluyo (Research Center for Vaccine and Drug, BRIN)
Anto Satriyo Nugroho (Artificial Intelligence and Cyber Security Research Center (PRKAKS), BRIN)

Jurnal Bioteknologi & Biosains Indonesia (JBBI)
Vol. 9 No. 2 (2022): December 2022

ABSTRACT Due to widely varied microscopic shapes, fungal classification can be performed based on their morphological features. In morphology-based identification process, feature extraction takes an important role to characterize each fungal type. Previous studies used feature extraction of fungal images to detect the presence of fungal. In this study, morphological and textural features were extracted to classify three types of fungi: Aspergillus, Cladosporium and Trichoderma. Geometry and moment were used as morphological features. To perform textural feature extraction, the local binary pattern (LBP) and gray level co-occurrence matrix (GLCM) feature extraction method were used. We compared the implemented feature extraction methods in order to get the best classification result. The result showed that geometrical features has the accuracy of 65%, higher than that of LBP (60%), GLCM (45%), and moment accuracy (55%). This suggested that geometric features is important for fungal classification based on their morphology.   ABSTRAK Karena bentuk mikroskopisnya yang sangat bervariasi, klasifikasi jamur dapat dilakukan berdasarkan ciri morfologisnya. Dalam proses identifikasi berbasis morfologi, ekstraksi ciri berperan penting untuk mengkarakterisasi setiap jenis jamur. Penelitian-penelitian yang dilakukan sebelumnya melakukan ekstraksi ciri citra jamur untuk mendeteksi keberadaan jamur. Dalam penelitian ini, fitur morfologi dan tekstur diekstraksi untuk mengklasifikasikan tiga jenis jamur: Aspergillus, Cladosporium dan Trichoderma. Geometri dan momen digunakan sebagai ciri morfologi. Untuk melakukan ekstraksi ciri tekstur, digunakan metode ekstraksi ciri local binary pattern (LBP) dan gray level co-occurrence matrix (GLCM). Kami membandingkan metode ekstraksi fitur yang diterapkan untuk mendapatkan hasil klasifikasi terbaik. Hasil penelitian menunjukkan bahwa fitur geometri memiliki akurasi 65%, lebih tinggi dari LBP (60%), GLCM (45%), dan akurasi momen (55%). Ini menunjukkan bahwa fitur geometris penting untuk klasifikasi jamur berdasarkan morfologinya.