Design and Development of Hyper Spectral Image Classification Using Enhanced Mean Shift Segmentation in Image Mining

Hyper spectral imaging is becoming an important analytical tool for generating land-use map. High
dimensionality in hyper spectral remote sensing data can guaranty in principle a detailed discrimination of
the observed surfaces overcoming the intrinsic limitation of lower spectral resolution data. We propose an
ovel approach for solving the perceptual grouping problem in vision. In the existing methods the image
segmentation was done by using special spectral classification. Special spectral may have some of the
problems such as quality and clarity of the particular image. We show that an efficient computational
technique based on a generalized region value problem can be used to optimize this criterion. In our
proposed method Enhanced mean shift algorithm is used for segmenting the part of the particular image.
Experimental result use accuracy and execution time parameters to show the performance. It takes low
computational complexity and also accurate result for real-time image segmentation processing