Investigating ISODATA Misclassification in Semi-urban Area

Semi-urban area forms its own ‘landscape' with low density, apparently random, scattered or fragmented and leap fogging forms of urban land use. The area under investigation is the Arasikere Semi-urban Area, located at 44km North of Hassan District, Karnataka State, INDIA with an elevation of approximately 806 m (2,644 ft) Above Mean Sea Level and is known for its coconut production. The data are of LISS-IV (Linear Imaging and Self Scanning) sensor of IRS-P6 (Indian Remote Sensing Satellite) and Panchromatic image of IRS-P5 satellites launched and maintained by the Indian Space Research Organization (ISRO). The images are characterized by many noises such as, clouds, haze, roofs and roads covered by rainwater, which cause confusion between urban class, water body and wetland. Hard classification was applied with ISODATA unsupervised classification technique and result is a proof of good choice of the study area characterized with mixed classes. Hard classification is a good tool for homogeneous area where no mixed pixels exist. ISODATA hard classifier failed to classify heterogeneous nature of Arasikere Semi-urban area.