Bmnd improves denoising for scientific data across all dimensions
BMND: Direct Poisson Denoising by N-Dimensional Block Matching and Collaborative Filtering
Computer Vision and Pattern Recognition
Summary
Cleaning up noisy scientific data is tricky because the noise depends on the signal itself and data come in many shapes and sizes. The authors created BMND, a method that directly handles this signal-dependent noise without changing the data first, working on 1D signals, 2D images, and 3D volumes alike. Their approach carefully matches patterns in the data while considering noise levels and keeps the total measured intensity accurate. Tests showed BMND improves clarity and preserves important details better than previous methods. The method is available as free software for anyone to use.
What this means in practice
- •For medical imaging technicians: Enhance clarity of 3D medical scans by reducing Poisson noise while preserving signal intensity using BMND.
- •For microscopy imaging engineers: Improve fluorescence microscope image reconstruction by directly denoising Poisson noise with BMND’s noise-aware matching and filtering.
Authors
Christof Duhme, Lars Schiefelbein, Florian Büther, Xiaoyi Jiang
Abstract
Poisson denoising of scientific data requires methods that account for signal-dependent noise while accommodating different data dimensionalities and preserving quantitative intensity information. We present BMND, a dimension-independent extension of block matching and collaborative filtering for Gaussian and Poisson observations. Building on the two-stage structure of BM3D and BM4D, BMND processes Poisson data directly, without a variance-stabilizing transform, by combining noise-aware patch matching with propagation of signal-dependent noise variances through collaborative filtering and aggregation. A dimension-independent reference-patch traversal scheme supports arrays with an arbitrary number of axes. An optional aggregation-aware mass conservation preserves the observed total intensity after weighted overlap-add. We evaluate the framework on one-dimensional physiological signals, two-dimensional images, and three-dimensional volumes, using controlled noise experiments and measured fluorescence microscopy acquisitions. The experiments demonstrate improved reconstruction quality from noise-aware matching and Wiener filtering, while low-count phantom experiments show reduced denoising-induced intensity loss through mass conservation. The framework provides a unified, non-learning-based approach to denoising across arbitrary data dimensions and is released as an open-source library.