Locally Linear Image Structural Embedding for Image Structure Manifold Learning

TL;DR

This paper introduces LLISE, a novel manifold learning method that uses SSIM instead of MSE to better capture image structure features and assess image quality.

Contribution

The paper proposes LLISE, a new manifold learning approach that incorporates SSIM for improved image structure representation and distortion discrimination.

Findings

LLISE effectively captures image structure features.

Kernel LLISE enhances manifold learning for image quality assessment.

The method bridges manifold learning with image fidelity evaluation.

Abstract

Most of existing manifold learning methods rely on Mean Squared Error (MSE) or $\ell_2$ norm. However, for the problem of image quality assessment, these are not promising measure. In this paper, we introduce the concept of an image structure manifold which captures image structure features and discriminates image distortions. We propose a new manifold learning method, Locally Linear Image Structural Embedding (LLISE), and kernel LLISE for learning this manifold. The LLISE is inspired by Locally Linear Embedding (LLE) but uses SSIM rather than MSE. This paper builds a bridge between manifold learning and image fidelity assessment and it can open a new area for future investigations.