Land is difficult for the administrator to make decisions

Land use change is happening very fast andconventional valuationmethods are unable to determine the current development patternefficiently. With fasterdevelopment pattern, it requires a system design and more efficient development for efficient controls. The weakness in spatial analysis in making decisions is less accurate and takes a long time.

The interpretation of the resultinganalysis is also difficult to do. Thereforeit is difficult for the administratorto make decisionsquickly and clearly.GIShelps to create space inventory on natural resources and environmentalconditions that play an important role in the spatial and dynamic evaluation ofthe area (Mondal et al., 2015). Additionally,Lo and Choi (2004) state that the combination of GIS technology with high spatial resolution satelliteimagery and advanced image processing has enabled monitoring and modeling ofmore routine and consistent to detect land use patterns.GIS  development is an  opportunity  that  must be taken  to  replace  the conventional system in the management of natural resources. Chillar (2000) indicated that GIS has abilityfor digital data storage that make manipulation and analysis process can be done quicklyandaccurately.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

The analysis results are easyto interpret and translate in the real world(Olokeogun et al., 2014). The advantage ofcreating a system ofmanagementfor natural resources will be moreefficient.2.

8       LULCImage ClassificationClassificationtechnique are commonly used to extract the LULC information based on thespectral signatures derived from the remote sensing images. Usually,pixel-based classification have been used to classify the remote sensing data.Pixel-based classification techniques is divided into two main categories whichare supervised classification and unsupervised classification. Unfortunately,this pixel-based classification has limitation especially for high resolutiondata (Xiaoxia, 2000). Therefore, object-oriented classification approach wasused to overcome the limitation. 2.8.1   ObjectOriented ClassifierThe rural and urbanareas have a complex spatial environment because typically, the environmentconsist of many build up structures, various vegetation covers, bare soil andwater bodies.

Therefore, improved image analysis techniques are required. Hayand Castilla (2006) have proposed Object-Based Image Analysis (OBIA) asappropriate method to classify the remote sensing data. Object-orientedanalysis extract information from remote sensing imagery and turns them intomeaningful image objects based on object texture, shape and contextual relationshipwith other objects.Object-oriented classificationis based on image segmentation by considering each pixel as a separate objectand may results in a more homogeneous and more accurate mapping product withthe higher details in class definition (Willhauck, 2000; Herold and Scepan,2002).

Segmentation principally means the adjacent pairs of image objects aremerged to form bigger segments. The merging decision is based on the localhomogeneity criterion. The homogeneity criterion is a combination of colour(spectral values), shape properties, smoothness and compactness (Willhauck,2000).eCognition isthe first object-oriented image analysis software on the market (UserGuidee-cognition,2003). It was produced by Definiens.

Segmentation in eCognition(Baatz and Schape, 2009) allows both segmentation based on primary features(gray tone and shape) and after an initial classification the more advancedclassification based segmentation. eCognition software supports the import andexport data of classification results in shape format. Since eCognition is aregion based analysis system, only polygon can be imported and use in other GISsoftware such as ArcGIS. Hierarchical network creation in eCognition like the one displayedin Figure 2.5 shows that objects created on different scale can be linkedtogether. The Multiresolution segmentation algorithm based on the Fractal NetEvolution Approach (Baatz and Schape, 2000) followed by Spectral differenceSegmentation algorithm (Trimble, 2014) were conducted to the data, in the form ofmultilevel hierarchy approach.