A Hybrid Shadow Removal Algorithm for Vehicle Classification in Traffic Surveillance System

摘要

Shadow is one of the common parts in the natural scenes and has become an important topic in the field of computer vision. In many vision-based traffic surveillance systems, shadows interfere with fundamental tasks such as vehicle detection, classification, and tracking. Thus, it is necessary to suppress the effect of shadows. A difficult part of the shadow removal problem is how to accurately detect and remove shadow regions and recover the boundaries of the vehicles, while still achieving real-time processing performance. Many powerful methods have been proposed to solve this dilemma; however, instabilities at the boundaries of moving vehicles are still challenges. In this paper, an improved algorithm to remove shadow regions, and quickly recovering the boundaries of moving vehicles is presented in a detailed manner. The proposed method applies edge information with background subtraction to handle daytime traffic scenes. Our approach has demonstrated more accurate results than previous approaches regardless of lighting luminance levels or shadow orientations.