ABSTRACT
There has been a significant growth in the volume and quality of remote sensing data over the last 25 years. Spatial and spectral resolutions have improved greatly. Satellites specially designed for dedicated purposes have been launched. High spatial resolution has been achieved at the cost of image quality in some cases. These have called for special image restoration algorithm development. Similarly, this high spatial resolution has brought greater importance to texture as an element and hence texture based classification methods have come into vogue. Likewise, shape also plays a key role and hence shape based Pattern Recognition (PR) tools have become important. The need to get better classification accuracy when the class boundaries are highly nonlinear has prompted the use of models such as Artificial Neural Networks (ANN) (these are mathematical models which mimic the human neurons and are used in many classification/learning applications). In the area of Digital Image Processing, apart from basic tools which manipulate the brightness, contrast or gamma, the main emphasis has been on image restoration, which models the sensor, platform and the atmosphere as well as on image models which lead to image understanding and simulation.
