Bio-SFT: Asymmetric Cortical Guidance and Retinal Adaptation for Robust HDR Reconstruction

2026-07-19

AI Summary

This work addresses the highly ill-posed problem of reconstructing high dynamic range (HDR) images from a single standard dynamic range (SDR) input, which often suffers from artifacts and color distortions—particularly in dark regions. To this end, the authors propose Bio-SFT, a biologically inspired spiking frequency transformer that integrates a learnable Naka-Rushton retinal adaptation mechanism, an asymmetric Parvo-to-Magno pathway guidance strategy, and a hard-gated spiking neural network (SNN) noise suppression module based on all-or-none spike encoding. Coupled with sparse prior training and a Transformer backbone, the proposed method achieves significant improvements in HDR-VDP-3 and ΔE_ITP metrics on the HDRTV1K dataset, effectively suppressing dark-region artifacts while preserving structural fidelity and enhancing perceptual quality in HDR reconstruction.

Abstract

Recovering high dynamic range (HDR) radiance from a single standard dynamic range (SDR) image is highly ill-posed. Extreme luminance variation and severe quantization in dark regions make accurate reconstruction challenging, often leading to visual artifacts and color distortions. To address this problem, we propose Bio-SFT, a bio-inspired spiking frequency transformer for single-image HDR reconstruction. Bio-SFT incorporates three biologically motivated components. First, a learnable Naka--Rushton retinal adaptation frontend stabilizes the input under complex lighting conditions. Second, an explicit Parvo--Magno split introduces asymmetric Parvo-to-Magno guidance, allowing high-frequency structural cues to modulate low-frequency reconstruction. Third, an event-driven SNN hard gating module applies all-or-none spiking to suppress dark-region noise while preserving structural details. The module is trained with a sparsity prior to encourage efficient feature utilization. Built for end-to-end training within a transformer backbone, these lightweight components provide strong parameter efficiency. Experiments on HDRTV1K show that Bio-SFT achieves competitive perceptual quality and consistently improves HDR-VDP-3 and $ΔE_{ITP}$ while reducing artifact propagation in symmetric guidance pipelines.

Problem

• HDR reconstruction
• single-image
• visual artifacts
• color distortions
• dynamic range

Innovation

• bio-inspired
• spiking neural network
• retinal adaptation
• asymmetric guidance
• HDR reconstruction