Research And Application Of Fire Source Location Technology Based On The Image In Spacious Buildings

With the development of computer vision and digital image processing technology,image-based fire locationing technology has become a hot research spot in recent years.In order to solve the problem that low accuracy and poor real-time performance of fire localization results in traditional image-based fire locationing technology,this paper researched an improved Genetic Algorithm solving framework based on flame image entropy speed up robust features(SURF, Speeded Up Robust Features) which is applied in the space building fire source localization.(1) Extraction and match of the flame image feature points. Research the technology that flame image feature points extraction and matching, validate and comparative analysis several feature point extraction and matching methods which are being existed and widely used, Because the feature points extraction effect of the SURF algorithm for brightness change image is better than other algorithms, it is used to extract the flame feature points.Then in order to get a much better matching points and fully preparation for behind the locationing, feature points entropy is introduced to reduce the error match.(2) Fire source camera calibration. Research the three kinds of existing calibration method, and combined with the fire source characteristics in large space building, the camera self-calibration method for calibration is selected. Through improving the selection, crossover and mutation operator to design genetic algorithm to solve the camera intrinsic parameters, which is suitable for camera calibration in large space building fire source location,and then use position relationship among binocular cameras to determine the external parameters.Finally, use triangular positioningprinciple to calculate three-dimensional location of fire source in large space building,so that complete the location.The experimental verification of the algorithm put forward by the paper has been done. The results show that in the part of the flame image feature point extraction and matching, the mismatch significantly reduced, and the problem of time delay can be solved; In the fire source camera calibration section, compared with other existing methods, the camera intrinsic parameters precision is higher than others; In the final 3D depth information calculating section, multiple ranges are test. Compared with the original method, the value is closer to the actual value, and the accuracy is higher.