← Back to all news
arXiv Computer Vision

The Skin-Restricted Reinhard Transform:Uniqueness under a Lightness-Preserving Constraint

The paper introduces the skin‑restricted Reinhard transform, a diagonal affine mapping in CIE Lab that preserves lightness while adjusting chromaticity for skin recolouring. It fixes the lightness gain to +1 and the shift to the mean difference, then optimises chromatic gains within [0.72, 1.18] using quadratic transport and projection. Experiments on nine images show the method maintains a lightness contrast ratio of 0.974 ± 0.029 with only 0.77 CIE Lab chromatic error, outperforming traditional Reinhard, Monge, and histogram matching approaches that reduce lightness contrast.

Related stories

arXiv Computer Vision
Sep 17

Geometry-Driven Shadow Harmonisation for Composited Faces: A Multiplicative, Albedo-Preserving Relighting Pipeline

The paper introduces a geometry‑driven shadow harmonisation pipeline for composited faces that preserves albedo by applying a per‑pixel gain field derived from a rasterised 3D face proxy. The method estimates key‑light direction from host cues, computes surface normals and cast shadows, and applies a controlled, monotonic darkening operation that never alters chromaticity or identity. Experiments on analytic face heightfields show that the default parameters modify over half the face pixels while maintaining hue invariance and avoiding repainting the donor face.

By Vijesh KP
More like this →
arXiv AI
Aug 26

Fidelity Preference, Not Demographic Preference: A Pixel-Level Attribute-Sensitivity Audit of Image Aesthetic/Preference Scorers

The study audits four image aesthetic scorers—LAION-Aesthetics, PickScore, ImageReward, and HPSv2—using pixel‑level interventions on skin tone and body type in both synthetic and real images. It finds that most scorers exhibit a fidelity preference: unaltered images receive the highest scores, while perturbations in either direction are penalized in an inverted‑U pattern, and this effect is largely independent of the skin operator. Synthetic‑only audits are misleading, as the apparent preference for darker skin in synthetic faces reverses or weakens when evaluated on real faces, and cross‑scorer results vary widely, underscoring the need for real‑data, within‑image causal isolation to accurately assess demographic bias.

By Mingyang Xu
diffusionbenchmarkssafety
More like this →
arXiv Computer Vision
Sep 4

Off the Planckian Locus: Using 2D Chromaticity to Improve In-Camera Color

The paper proposes replacing traditional one-dimensional correlated color temperature (CCT) interpolation with a two-dimensional chromaticity approach for in-camera colorimetric mapping. By training a lightweight multi-layer perceptron (MLP) on a lightbox calibration that includes representative LED sources, the method achieves a 22% average reduction in angular reproduction error across diverse LED lighting. It remains compatible with conventional illuminants, handles multi-illuminant scenes, and can be deployed in real time with minimal computational overhead.

By SaiKiran Tedla, Joshua E. Little, Hakki Can Karaimer, Michael S. Brown
More like this →
arXiv AI
Aug 18

Picking the Right Image to Classify: Reliable-Input Selection in Teledermatology

arXiv:2608. 16198v1 Announce Type: cross Abstract: Dermatology models face distribution shifts in teledermatology settings, where submitted images differ from the training data in lighting, angle, distance, focus, and framing.

By Fabian Gr\"oger, Marco Weishaupt, Philippe Gottfrois, Simone Lionetti, Linda Wermelinger, Nipun Ranasekara, Ludovic Amruthalingam, Alexander A. Navarini, Marc Pouly
ragbenchmarks
More like this →