ANALISIS MATHEMATIK FRAKTAL UNTUK KLASIFIKASI MENGGUNAKAN CITRA PENGINDERAAN JAUH SPOT-4

Abstract

Fractal is a mathematical set that typically displays self-similar patterns. Fractal have two basic characteristic suitable for modeling the topography of the earth surface self similarity and randomness; Applications of fractal geometry in remote sensing rely heavily on estimates of the non integer fractal dimension (D). The fractal dimension is calculated using the model of Surface Area Triangular Prism (TPSA). Fractal dimension is used to observe the spatial repetition (morphologie) of surface. In this study, fractal dimension is used to observe the relative height of a building / object of surface in urban area. This paper described image analysis using non integer fractal dimension used to determining the height of an object relative to the others, then do grouping of the object height by thresholding method. The result of the whole proses is presented after the density slicing proses. The analysis showed that the fractal dimension of the homogeneous object/surface is smaller than the heterogeneous objects. Based on it’s fractal dimensional objects/buildings in Jakarta city (covering 1600 ha), can be grouped in 3 classes: very high object, high object and rather high object and there are approximately 178 ha using 9 x 9 windows and approximately 80 ha using 17 x 17 windows very high object. However, the results of this study are still in the early stages that the fractal dimension can quantitatively interprets spatial structure and spatial complexity of remote sensing data. Therefore, research needs to be followed up with the field measurements and very high resolution resolution data (such as IKONOS).