A Clustering Based Method for Edge Detection in Hyperspectral Images

Abstract

Edge detection in hyperspectral images is an intrinsically difficult problem as the gray value intensity images related to single spectral bands may show different edges. The few existing approaches are either based on a straight forward combining of these individual edge images, or on finding the outliers in a region segmentation. As an alternative, we propose a clustering of all image pixels in a feature space constructed by the spatial gradients in the spectral bands. An initial comparative study shows the differences and properties of these approaches and makes clear that the proposal has interesting properties that should be studied further.

Cluster-Based Spectral-Spatial Segmentation of Hyperspectral Imagery

“…Enhancements to these basic techniques to overcome known drawbacks or tailor the method to particular scenarios are also common [2], [10]. Per [3], common spatial information extraction methods include morphological profiling [11], edge detection [12], and Markov random field modeling [6], [13]. Techniques for fusion of spatial information include, but are not limited to: direct projection of the spectral cluster labels directly onto spatial locations [2], plurality voting (with or without regularization) [14], and support vector machines with composite kernels or stacked vectors [15], [16].…”

“…Edge-based segmentation approaches generally seek large gradients (i.e., rapid changes in the image) as a way of defining boundaries between isolated regions, although the exact meaning of "large" and "gradient" varies among the different approaches [3], [4], [9], [12], [17].…”