Performance evaluation of Hyperspectral image segmentation implemented by recombination of PCT and Bilateral filter based fused images

This paper proposes a novel method in which principal component transformed (PCT) images and bilateral filtering based fused images have been recombined to improve unsupervised hyperspectral image segmentation. PCT and bilinear filtering based approach have been used for visualization enhancement and segmentation. First three PCT images contain more than 95% energy and highest variance of the data. Bilinear filter preserves sharp edges and smoothen the image. Combined image will have advantage of both techniques and able to provide improved segmentation results. Experiments have been conducted on HYDICE hyperspectral image. For evaluation, we compared the segmentation results of the approaches against a ground truth dataset by various segmentation measures like Sensitivity, Precision, False Positive Rate (FPR), Accuracy and Matthews Correlation Coefficient (MCC). Proposed technique gives better performance than other technique and it has also been proved by the comparative results.