Tracking Algorithm of Multiple Pedestrians Based on Particle Filters in Video Sequences.

Area of Science:

• Computer Vision
• Artificial Intelligence

Background:

• Pedestrian tracking is vital in computer vision but challenging in complex environments due to posture changes, occlusion, and background motion.
• Particle filters are effective for nonlinear, non-Gaussian tracking but require enhancements for multi-pedestrian scenarios.

Purpose of the Study:

• To develop an improved particle filter-based algorithm for robust multiple pedestrian tracking in video sequences.
• To address challenges like occlusion, scale variation, and background interference.

Main Methods:

• Utilizes a priori knowledge to extract confidence values for object and background detection.
• Integrates color and texture features into particle filters, adaptively adjusting feature weights.
• Implements occlusion handling and a method for detecting object disappearance and emergence.

Main Results:

• The algorithm demonstrates improved detection and observation capabilities by leveraging adaptive feature weighting.
• Effective management of severe occlusion prevents tracking drift and loss.
• Robust tracking of multiple pedestrians is achieved, including handling appearance and disappearance.

Conclusions:

• The proposed algorithm significantly enhances multiple pedestrian tracking performance in complex video sequences.
• The adaptive feature integration and occlusion handling contribute to superior tracking accuracy and robustness.