Robust vehicle detection in different weather conditions: Using MIPM.

Area of Science:

• Computer Vision
• Intelligent Transportation Systems (ITS)

Background:

• Conventional image-based traffic detection struggles with perspective, noise, and environmental factors.
• Accurate vehicle segmentation and tracking are crucial for effective Intelligent Transportation Systems (ITS).

Purpose of the Study:

• To propose a novel method for robust vehicle segmentation and tracking in traffic surveillance.
• To improve the accuracy of vehicle detection under diverse and challenging conditions.

Main Methods:

• Modified Inverse Perspective Mapping (MIPM) for perspective correction.
• Hough transform for road line and lane extraction.
• Gaussian Mixture Models (GMM) for moving object segmentation and a chromacity-based strategy for shadow removal.

Main Results:

• The proposed method demonstrates robust performance across various video benchmarks, including urban and interurban settings.
• Achieved higher accuracy compared to existing methods, particularly in scenarios with occlusions, lighting variations, and adverse weather.

Conclusions:

• The developed algorithms offer a significant improvement for vehicle detection and tracking in ITS.
• The method is effective in real-world traffic scenarios, enhancing the reliability of traffic monitoring systems.