Application Research Of Computer Vision In Early Forest Fire Detection

Currently Forest fire has become a global problem. Therefore, in order to reduce the loss of forest fire, all the countries in the world attaches great importance to forest fire detection technology. Fore preventing the fire to occur, forest fire detection has become the subject with the deep research. With the development of computer technology and digital image processing technology, a intelligence forest fire detection based on computer vision will be possible. In this dissertation, by the aid of digital image processing techniques, several early forest fire detection and recognition algorithms have been realized.The process of forest fire as well as the main physical phenomenon during with it, have been detailed discussed systematically in this dissertation. Based on this, we discuss the early forest fire smoke character and flame character. The law and characteristic property have been discovered and provide important basis for forest fire detection.The techniques of fire image enhancement , noise filtering, and the algorithms of maximum entropy threshold segmentation, region growth, Otsu threshold segmentation have been studied in this dissertation. Compared with the experiment results. Presents temporal-spatial segmentation and Background restoration segmentation, and the smoke foreground has been successfully extracted. On the base of integration of several segmenting algorithms , a new method has been put forward that can track the fire flame and obtain the fire target.Early forest fire image recognition has been discussed in the last pare of this dissertation. The fire smoke image recognition and the fire flame image recognition have been studied respectively. For smoke, we present a series recognition algorithms by use of shape geometry characteristic and temporal frequency characteristic of smoke. For flame, the features of fire flame such as edge variety, shape variety, the whole motion trend and flame flicker frequency have been analyzed in this dissertation. And the corresponding recognition algorithms have been given. Experiment results show that these algorithms can successfully detect and recognize the early forest fire.