Automatic Vehicle Detection during Nighttime Using Bright Pixel Segment with Spatial Temporal Technique

The paper proposes an effective Traffic surveillanc e system for detecting and tracking moving vehicles in nighttime traffic. It identifies vehicles by detecting and locating vehicle headlights and taillights using imag e segmentation and pattern analysis technique. By preprocessing noise is remov ed using median filter. Morphological operation is used to extract candidate headlight objects and then perform shape analysis. Template matching or pattern classification to find the paired headlight of moving vehicles. Salient points are used to represe nt local properties of image classification. It car ries information about image content. Gabor derivation is used for edge detectio n and feature extraction.