Land use /land cover classification based on Object-oriented technique and satellite image Case study: West Azerbaijan Provinces

Optimal natural resources management depended on reliable as well as up-to-dated data. For this mean, land use/cover maps are considered as an important source of information on the natural resources management. Nowadays, remote sensing users could be able to derive different land use/cover maps from satellite images by using specific interpretation techniques. Some characteristics of Satellite images such as: digitally format, production up-to dated data, wide viewing angle (swath width), multispectral as well as multi temporal and revisiting time of data acquisition with high speed on data transformation make those be considered as valuable information on the natural resources management. In this research, land use/cover map of study area has been produced based on digital interpretation of SPOT5 satellite image (2005) by Object Oriented method. Based on proposed methodology, some pre-Processing practices which involve geometric and radiometric correction were implemented by applying Pci Geomatica 9.1 software. Image processing was conducted on the next step based on Object Oriented method by applying eCogenation software. Followed step related to classifying of SPOT image based on proposed classes (18 classes: irrigated agriculture, dry farming area, bare soil, apricot orchards. Apple gardens, vineyards…). Finally, accuracy of classified image was assessed by fulfilling of error matrix and calculating of overall accuracy as well as Kappa Coefficient. Result of overall accuracy assessment was calculated 93.43%. It was confirmed the reliability of classification results. Final step was proposed to implement geo database into GIS environment for both illustration of produced map and comparing with former which was derived by pixel method. Results of this research showed, by Object-oriented method one be capable to produce land use/cover maps simultaneously with high accuracy and more classes.