Hyperspectral image segmentation for paint analysis

Abstract

Hyperspectral imaging (HSI) is used in analysis of paintings to obtain features hidden to the human eye by selecting spe- cific wavelengths. Superpixel segmentation can be applied to HSI for feature extraction. A superpixel algorithm pro- cesses an image in a way in which the result includes an un- necessary amount of over-segmentation. In this work, we use over-segmentation and propose Spectral Similarity Merging (SSM), a region growing algorithm based on homogeneous spectral properties with the aim to reduce over-segmentation without compromising under-segmentation. The algorithm focuses on the similarity of the spectral shapes rather than intensity. Results show an average of 45% reduction in over- segmentation and an average of 53% improvement on the F- score on existing superpixel segmentation algorithms.