jack925/Newtons_rings

Newton's Rings Simulation Dataset

A synthetic dataset for Newton's Rings interference pattern analysis, designed for machine learning and deep learning research in optical physics.

Dataset Overview

• Task: Predict curvature radius R and ring center coordinates (x₀, y₀) from interference images
• Image Size: 512 × 512 pixels
• Image Format: PNG (8-bit grayscale)
• Label Format: JSONL (each line contains complete physical/imaging/noise parameters)

Dataset Configurations

This dataset includes multiple illumination mode configurations:

Configuration	Light Mode	Description
vis512_monochrome	Monochromatic	Fixed wavelength illumination
vis512_narrowband	Narrowband	Random narrowband wavelength
vis512_white	White light	Broadband illumination
vis512_real_lambda	Real wavelength	Common laser wavelengths
vis512_ood	OOD test	Out-of-distribution generalization test

Label Description

Each sample label contains:

• Physical Parameters: Wavelength λ, curvature radius R, refractive index n, initial gap t₀, etc.
• Imaging Parameters: Ring center coordinates, PSF blur, vignetting, etc.
• Noise Parameters: Poisson-Gaussian noise, read noise, dark current offset, etc.

Evaluation Metrics

Task	Metric
Curvature Radius Prediction (T1)	Relative Error |R̂ - R| / R
Center Localization (T2)	Euclidean Distance √[(x̂₀ - x₀)² + (ŷ₀ - y₀)²]

Usage

from datasets import load_dataset

dataset = load_dataset("jack925/Newtons_rings")

License Apache License 2.0 Citation If you use this dataset, please cite the associated paper. Contact School of Science, TianJin University of Commerce, AIboat project