Perbandingan Klasifikasi Tutupan Lahan Menggunakan Metode Klasifikasi Berbasis Objek dan Klasifikasi Berbasis Piksel pada Citra Resolusi Tinggi dan Menengah

摘要

A region will experience growth that it will bring changes in the physical appearance. Evolving region need to review land use planning to steer land cover allocation properly. It requires an accurate and effective method to obtain land cover information. One effective technology for mapping land cover is a remote sensing technology. There are various kinds of data processing techniques in remote sensing to obtain land cover information. Classification techniques in remote sensing image are divided into three parts classification technique that are pixel based technique, sub-pixel based technique, and object-based techniques. In this study, the pixel based classification and object based classification techniques will be compared in land cover classification on high resolution imagery that are Quickbird imagery dan medium resolution imagery that are Landsat 8 imagery with the location in city of Semarang. Comparison of the results object based classification and the pixel based classification is tested for accuracy by confusion matrix that produce land cover classification accuracy of Landsat 8 obtained value the overall accuracy for an object based classification method amounted to 77.14%, while the pixel based classification methods obtained a value of 75.71%. For Quickbird image, object based classification produce in overall accuracy of 87.14% while the pixel-based classification obtained a value of 82.85%. The results showing the accuracy of the object based classification is quite good compared to the pixel-based classification either at medium resolution imagery (Landsat 8) and high resolution imagery (Quickbird).