Efficient resolution of Mixed Pixels using bio-inspired heuristics

In recent years, several nature inspired algorithms have been applied for remote sensing image classification. Each pixel is categorized into one of the several land cover classes such as water, vegetation etc. When a pixel is categorized into more than one class at a time then such pixels are called Mixed Pixels. Existence of mixed pixels lowers classification accuracy of satellite image. This paper focuses on resolving the Mixed Pixels, using sinusoidal migration model in Biogeography based optimization (BBO). The methodology proposed also involves excluding the redundant bands from multiband image to provide better decomposition of Mixed Pixels.