GIS can hold agricultural regions data like forest, fruit covered lands and/or cultivate lands, these lands have been managed inside GIS by receiving a selected region remotely sensed image, so GIS users must have an appropriate digital map that represents theses lands each one according to its owner, status, and some other data. Normally, in such system, these lands will be classified by the users according to agricultural status depending on human vision. So, hardly to the users to classify these lands manually, and this become a great problem which take a long time depending on human efforts, especially if there is a huge number of lands. The suggested study creates a new ArcMap GIS tool which classifies these given lands automatically. Thus, this tool runs the developed system application; it will gather required information for each one of selected land, by sampling sub-images from their centers depending on the digital map, and gathers related status information from attribute database. On the next stage, the system will extract a vector of textural features for each one of the selected lands from their image samples using second order statistics Gray Level Co-occurrence Matrix (GLCM) and calculate eight textural features for each one of three visible bands (RGB) for each land sample. That vector of features will become the input to supervised multi-layer perceptron with backpropagation neural network classifier which be learneddepending on recommended GIS user training data set. As a result the system has accuracy near to 75%; these results were achieved by comparing the classification results from system test trials with desired user classification data.

---

 

Introduction:


Geospatial data has both spatial and thematic components. Conceptually, geographic data can be broken up in two elements: observation or entity and attribute or variable. GIS have to be able to manage both elements. Spatial component, the observations have two aspects in its localization, absolute localization based in a coordinates system and topological relationship referred. A GIS is able to manage both while computer assisted cartography packages only manage the absolute one. The aim of classification is to link each object or pixel in the study area to one or more elements of a user-defined label set, so that the radiometric information contained in the image is converted to thematic information, The process can be regarded as a mapping function that constructs a linkage between the raw data and the user-defined label set. Two types of classification are commonly performed. Supervised classification methods which are based upon prior knowledge of the number and certain aspects of the statistical nature of the spectral classes with which the pixels making up an image are to be identified, and unsupervised classification methods which are performed by running a classification algorithm without any predefinition of spectral classes of interest. Texture is also an innate property of objects. It contains important information about the structural arrangement of surfaces. The use of texture in addition to spectral features for image classification might be expected to result in some level of accuracy improvement, depending on the spatial resolution of the sensor and the size of the homogeneous area being classified. R. Methre et al. [38] investigated the texture retrieval using combination of local features of Haralick derived from one level discrete wavelet transform coefficients and global statistical features computed from three level wavelet transformed images. Y. Zhang et al. [44] proposed a hybrid classifier for polarimetric SAR images, the feature sets consist of span image, the H/A/? decomposition algorithm, and the GLCM-based texture features. Then, a probabilistic neural network (PNN) was adopted for classification. E. S. Flores et al [11] used GIS techniques to improve the classification capabilities of a feature extraction algorithm for land use/cover change detection in a deciduous forest environment.