Residential Area Recognition Using Oscillatory Correlation Segmentation of Hyperspectral Imagery

An extended oscillatory correlation segmentation algorithm is complied to perform unsupervised scene segmentation for hyperspectral imagery(HSI). According to perceptual mechanism, the high variances associated with bright intensity values are just salient regions of scene. Instead of lateral potential, saliency map is hired to obtain self-excitable cell. Then hyperspectral imagery is segmentated by extended LEGION. With these steps, more accurate initial residential areas can be obtained, but with many deficiencies including the existence of holes and useless patches. To resolve these problems, a morphological space based method is used to dissolve these residential patches. Experiment on PHI-3 data demonstrates the utility of the algorithm for residential areas recognition.