Fire Source Location Identified Based On Video In Fire Control System For Large Space Building

In the rapid development of the current society, large space buildings are becoming more and more. These places are crowded and have many types of internal goods and their structures are complex. The traditional smoke temperature and spray type fire system have lots of defects, they are not very good to achieve fire extinguishing. With the development of technology such as video camera, computer and so on, the fire control system can overcome the faults of the traditional methods, which is very suitable for large space buildings fire prevention. Video information is more intuitive, and contain full scale output of the fire scene information, at the same time it can carry out three-dimensional positioning of fire.Fire detection system based on video contains four parts: fire detection, image feature matching, fire location and intelligent fire gun. This thesis mainly studies the problem of fire location identify which obtain relevant parameters at the scene of the fire of the camera self-calibration, and establish the mapping relationship between the 2D image planes and three-dimensional spaces, located the three dimensional coordinate information of fire and ready for the follow-up of intelligent fire gun extinguishing. This thesis mainly has the following three aspects:Firstly, this thesis use SURF algorithm to extract the image feature points, the method has rotation, translation and scale invariance and faster characteristics. In the matching part, with a fusion method based on ratio method and the correlation function method, that advanced line rough matching, and then the study make its calculation speed and get a more accurate matching point with good robustness.Secondly, the thesis study the calibration technique of the intersecting binocular camera. That propose an improved based adaptive differential evolutionary algorithm of camera self-calibration method based on Kruppa equations self-calibration technique and introduce the simulated annealing algorithm in order to realize the camera calibration. In the end, the more accurate internal and external parameters of the camera were obtained.