Research On Video Recognition Technology Of Forest Fire

The forest fire is one of the most devastating natural disasters, countries are working on forest fire monitoring techniques, traditional monitoring methods due to the cost of large, low efficiency, monitoring work is very difficult. With the development of video surveillance equipment and digital image processing technology, the fire automatic monitoring system based on machine vision emerged, It has the advantages of low cost, testing large amount of information, and improve the response speed of early prevention measures.This article describes the process of early forest fire image characteristics and fire recognition system,focuses on the video-based fire detection algorithm, the combination of background subtraction method based on HSV color space to extract a moving target, Otsu threshold method to deal with a target processing, and morphological filtering noise, thus obtains a complete goals. Extract the suspected area eigenvalue calculation area growth rate, similarity, texture features, circularity. GVF snake model approach to edge detection of the fire area. Finally, design the BP neural network system to identify fire.Experiments prove that the forest fire recognition system based on video recognition rate is higher.