Review of wavelet-based unsupervised texture segmentation, advantage of adaptive wavelets

TL;DR

This paper evaluates the impact of different wavelets on texture segmentation and demonstrates that adaptive empirical wavelets improve segmentation accuracy, especially when combined with a cartoon-texture decomposition step.

Contribution

It introduces the use of empirical wavelets for texture segmentation and shows their advantages over traditional wavelets in this context.

Findings

Empirical wavelets outperform classic wavelets in segmentation accuracy.

Preprocessing with cartoon + texture decomposition enhances segmentation results.

The method is validated on six standard texture benchmarks.

Abstract

Wavelet-based segmentation approaches are widely used for texture segmentation purposes because of their ability to characterize different textures. In this paper, we assess the influence of the chosen wavelet and propose to use the recently introduced empirical wavelets. We show that the adaptability of the empirical wavelet permits to reach better results than classic wavelets. In order to focus only on the textural information, we also propose to perform a cartoon + texture decomposition step before applying the segmentation algorithm. The proposed method is tested on six classic benchmarks, based on several popular texture images.