Related Experiment Video
Updated: Aug 16, 2025
08:18
High-Accuracy Correction of 3D Chromatic Shifts in the Age of Super-Resolution Biological Imaging Using Chromagnon
Published on: June 16, 2020
7.5K
Uniformity Correction of CMOS Image Sensor Modules for Machine Vision Cameras.
Gabor Szedo Becker1, Róbert Lovas2
1Doctoral School of Applied Informatics and Applied Mathematics, Óbuda University, Bécsi út 96/B, 1034 Budapest, Hungary.
Sensors (Basel, Switzerland)
|December 23, 2022
Summary
This study analyzes how temperature and gain affect image sensor nonuniformities (DSNU and PRNU) in machine vision cameras. It proposes hardware solutions for improved image uniformity and performance across various applications.
Area of Science:
- Image Signal Processing
- Machine Vision Systems
- Semiconductor Device Characterization
Background:
- Image sensor nonuniformities, including dark signal nonuniformity (DSNU) and photoresponse nonuniformity (PRNU), degrade image quality.
- These nonuniformities, along with lens system characteristics, are known to be temperature-dependent, impacting sensitive machine vision applications.
- Existing flat-field correction (FFC) methods often rely on multiple calibration images to address these variations.
Purpose of the Study:
- To characterize the temperature and analog gain dependence of DSNU and PRNU in global shutter CMOS image sensors.
- To propose and evaluate an optimized hardware architecture for compensating image sensor nonuniformities.
- To compare different nonuniformity correction approaches based on performance, cost, and manufacturability.
Main Methods:
- Characterization of DSNU and PRNU across varying temperatures and analog gain settings for two CMOS image sensors.
- Development of an optimized hardware architecture incorporating optional parametric lens shading correction (LSC).
- Evaluation of compensation strategies using single or multiple calibration images captured under different conditions.
Main Results:
- The study quantifies the temperature and gain dependency of DSNU and PRNU for the tested sensors.
- An optimized hardware architecture was proposed, offering different configurations for diverse application needs.
- Comparison of various correction approaches highlighted trade-offs in effectiveness, resource utilization, and complexity.
Conclusions:
- Temperature and analog gain significantly influence image sensor nonuniformities, necessitating adaptive correction strategies.
- The proposed hardware architecture provides a flexible solution for compensating nonuniformities in machine vision.
- The choice of correction method should be tailored to specific application requirements, balancing performance, cost, and implementation complexity.
