Kajian Pemanfaatan Jaringan Saraf Tiruan Untuk Klasifikasi Jenis Tanaman Pertanian Pada Foto Udara Format Kacil

• Main Author: Perpustakaan UGM, i-lib
• Format: Article NonPeerReviewed
• Published: [Yogyakarta] : Universitas Gadjah Mada 2006
• Subjects: Jurnal i-lib UGM
• Online Access: https://repository.ugm.ac.id/25478/; http://i-lib.ugm.ac.id/jurnal/download.php?dataId=8473
• Institution: Universitas Gadjah Mada

Generally, conventional supervised parameter classification methods show low performance if applied to high spatial resolution image, such as Small Format Aerial Photography (SFAP). Supervised parameter classification methods just involved spectral color in process of classification and the pixels must have normal distribution. Application the methods for high spatial resolution image (SFAP), in which the pixels distribution are not normal distributed, causing improper result. For this reason, manual classification using visual interpretation shows better result rather than supervised parameter classification methods. Unfortunatelly, visual interpretation needs a lot of operator interactions and can cause inconsistency results, depand on local knowledge of each operator. So, this research studied to classy agricultural plantation using artificial neural network (ANN) from SFAP. Method of ANN that used in this research is supervised multi layer perceptron with back propagation training. Result shows generalization number of class (from 9 classes to 6 classes and finally to 5 classes) based on JeffriesMatusita's separability index for each training area produces the best accuracy. Classification assessment shows overall accuracy is 84,28% and Kappa's coefficient is 0,78. Kata kunci:Teknik Sipil-Perencanaan .