Active Illumination-Assisted Multiview Camera System for Robust 3D Visual Perception Under Challenging Environments
CSEF · 2026 Electronics & Electromagnetics (Senior Division)
Overview
This project presents computational techniques and hardware architecture for robust 3D perception in challenging environments, where signal degradation causes conventional sensors to fail. The hardware utilizes active illumination multiview camera architecture. It comprises dual NIR cameras, an IR light-sheet module, and a calibrated galvo scanner synchronized through a high-speed DAQ card. By integrating geometric constraints and precise timing, the system operates as a gated sensor, filtering environmental noise to maximize the signal-to-noise ratio in degraded conditions. A novel spatio-temporal correlation algorithm was developed for accurate feature localization and subpixel matching. By leveraging information from both domains while simultaneously enforcing spatial correspondence and temporal coincidence, the dual-domain approach provides a significant advantage in maintaining robust recovery despite corrupted input. Additionally, this study identifies device specific signatures to mitigate adverse weather effects and introduces the concept of multi-echo active stereo for recovering multi-layered depth scenes. The system achieves a depth error below 1.5% at 30 m under clear conditions. Robustness was evaluated across three test categories: 1. Adverse weather conditions: dense fog, heavy rain, snow 2. Scenes with high geometric and material complexity 3. Detection of low-observability objects: transparent occluders, small-scale structures, multi-layered depths, and detection through high-scattering media The results demonstrate consistent 3D reconstruction with minimal degradation in depth accuracy. While recovery noise increases with the severity of environmental degradation such as fog density, the geometric integrity of the recovered scene remains intact. Potential applications of this study include autonomous navigation and robotics.
Competition history
- CSEF 2026
Related projects
ISEF · 2018
Seeing Clearly and Farther: Augmented Perception for Safe Driving in Adverse Weather and Reduced Visibility Conditions
ISEF · 2015
Computer Vision: Mapping and Orientation in 3-D Space
CSEF · 2017
Six Degree Positioning Based on Laser Plane Projection and Inertial Navigation
ISEF · 2015
Constructing a Single Camera Triangulation System for Capturing 3D Point Clouds with Color
CSEF · 2026
AI Nightguard Vision+
ISEF · 2021
3D Depth Mapping Vision System Worthy of an Extraterrestrial
ISEF · 2021
Poly-open LIDAR: Fast and High Resolution Adaptive LIDAR with original Multi-Opening Polygon Mirror System
ISEF · 2021
Fusing LiDAR and Camera Data for Advanced Context Recognition in Autonomous Navigation Sensory Systems Through Multidimensional Deep Neural Network Architectures
Closest projects by meaning, across every fair and year in the corpus.
Browse more like this
Source: California Science & Engineering Fair public projects