Remote sensing data provides much essential and critical information for monitoring many applications such as image fusion, change detection and land cover classification. This paper proposed about the classification and extraction of spatial features in urban areas for high resolution multispectral satellite image. Spectral information is the foundation of remotely sensed image classification. Initially, Preprocessing is done for multispectral satellite image using Gaussian filter. Then the features are extracted from the filtered image using Gray Level Co-occurrence Matrix (GLCM).Finally, Extracted features are classified using Back Propagation Artificial Neural Network (BPANN) and the performance is analyzed based on its accuracy, error rate and  sensitivity.