| Fire disaster is one of the great threats of the human security. Frequent wide area fire, forest fire, etc. cause not only great destruction to social life, but also the losses and the extent of injury. Fire detection and auto-put-out technology based on image processing is to use CCD camera as the probe, and make the color and ultra-red images in the watch scene input into the computer, then using the algorithms of image processing, to recognize whether there is fire occurring from images. If there is fire occurring, calculate the space position of the fire, and start the auto-put-out system and alarm system automatically. The thesis using the feature of fire flame morphologic and flame vision, researches the technology of fire detection and auto-put-out based on image processing, which adopt the BP neural network model. The aim is to design and realize the whole fire detection and auto-put-out system using software.First of all, on the technology of fire detection, the thesis finds out a series of algorithms for the recognition of flame shape features. For improving efficiency, the algorithm of region segmentation is brought forward, which makes the average processing time of the algorithm lessen to 1/4 comparing the classic algorithm. For lowering the complexity of computing the sharp angle, a new method based on boundary chain code is introduced. What's more, the Roundness of each enlightened areas is calculated, which makes better result.Then, on the technology of auto-put-out fire, the thesis researches the method of calculating fire position in color images. Also, based on advantage and disadvantage of universal binocular imaging positioning method and the method of rotating camera positioning, an algorithm of compound positioning is studied, which can improve the veracity of positioning.Finally, based on all have been researched, the thesis brings forward an overall design for the system. The experiment results show the system has the good feature of veracity and reliability. |
PDF Full Text Request