Related Experiment Video
Updated: Apr 23, 2026
12:39
A Methodology for Capturing Joint Visual Attention Using Mobile Eye-Trackers
Published on: January 18, 2020
7.0K
Vanishing point-based image transforms for enhancement of probabilistic occupancy map-based people localization
Summary
This study introduces novel image transforms for improved people localization in surveillance systems. These transforms enhance accuracy by ensuring upright person appearance, even with varied camera setups.
Area of Science:
- Computer Vision
- Robotics
- Artificial Intelligence
Background:
- Vision-based surveillance systems are prevalent, driving research in people localization.
- Probabilistic Occupancy Map (POM)-based methods are common for people localization.
- Camera configuration deviations can degrade localization accuracy.
Purpose of the Study:
- To propose novel image transforms for enhancing POM-based people localization.
- To improve localization accuracy under general camera configurations.
- To maintain computational efficiency in people localization schemes.
Main Methods:
- Developed image transforms utilizing the vanishing point of vertical lines.
- Transforms are based on image or ground plane coordinate systems.
- Aims to produce transformed images where people appear upright.
Main Results:
- The proposed transforms significantly improve POM-based people localization accuracy.
- Effectiveness is demonstrated for more general camera configurations.
- Localization accuracy degradation due to camera deviation is alleviated.
Conclusions:
- The novel image transforms enhance people localization in vision-based surveillance.
- The method offers improved accuracy and robustness across diverse camera setups.
- This approach maintains computational efficiency for practical applications.
