Scattered Image Dataset with Non-Coherent Light for Deep Learning Restoration

Published: 6 May 2025| Version 2 | DOI: 10.17632/nm233hnd6y.2
Contributor:
Roger Chiu

Description

This dataset consists of grayscale image pairs, where each pair includes a scattered input image and its corresponding ground truth projection. The data was acquired using a custom, low-cost optical setup built around a Raspberry Pi, which projects structured patterns through various optical diffusers and captures the resulting scattered light with a digital camera. All images are stored in PNG format with a resolution of 256×256 pixels and 8-bit depth. Data collection was performed under different scattering conditions, achieved by employing multiple diffusers with varying physical thicknesses and scattering coefficients. The projected patterns comprise digits, geometric figures, and textures, providing a diverse set of visual features. Each dataset entry thus offers a direct mapping between a scattered observation and the original undistorted pattern, supporting research on deep learning-based image restoration and scattering compensation.

Files

Institutions

  • Universidad de Guadalajara

Categories

Deep Learning, Deconvolutional Network

Licence