Forest Fire Monitoring Based On Video Image

This paper based on early phase forest fire monitoring, application of digital image processing technology and artificial neural network technology to monitor forest fires in the early identification of the research, and offers theoretical preparation and technology support to the forest fire intellectualized monitored in the future.The research way for this paper according with the basic step of the image recognition is that the first, study based on the video image capture card image collection, and the images collected by the mean and median filtering filter combination of pretreatment methods of filtering denoising. Secondly,introduction the difference method based on background moving target detection and background updated strategy in real-time, for the difference adoption of optimal threshold value method of moving target binary. On the target region's isolation, use the morphology of corrosion and expansion method eliminated, and thus gained complete moving target. The goal in the on going movement, so this with two frames of background difference images as a group to feature extraction, extracted two frames changes as a feature, and normalization of it. Finally,the adoption of improved BP neural network training samples, and the training error curve drawn, through the trained network sample tests, show that it's feasible of use neural network technology to achieve early identification for forest fires, and to identify high success rate, rapid identification, the accuracy and effectiveness. The system is the base for further research.