Research on Large Space Building Fire Positioning Technology Using Video Surveillance Images

Abstract

The large space building fire source locating technology is an important link to automatic fire detection and extinguishing. This paper introduces a method that determines fire source location using three-dimensional reconstruction techniques based on accurate fire detection. The process includes: feature extraction, feature matching, and 3D spatial locating. The feature extraction process improves the scale invariant feature transform (SIFT) algorithm. It reduces feature points’ 128-dimensional vector description to 16-dimension. The improved SIFT algorithm accelerates feature extraction while maintaining the good invariance of rotation, scale, and lamination factors from the original algorithm. To improve the positioning accuracy of the feature point matching process, this article adopts bidirectional matching method to achieve the exact match of image features according to the principle of unique match of feature points. In order to further improve the positioning accuracy, this article calculates the depth of the fire source from the scene by improving the binocular vision method and the use of the camera-cruise in determining different locations of a fire in the image. And finally it corrects the fire depth information by applying the confidence value. KeywordsSIFT algorithm; fire vide; parallel binocular vision