A Multi-feature Fusion Method Based On Video Fire Detection Is Proposed

The use of smoke, temperature of the traditional fire detection technologies exist poor adaptive capacity, such as lack of stability,can not be sustained, especially in the space of some tall buildings, the fire is too large, the complexity of environmental interference, it is difficult to carry out effectively. We at last formulated a multi-feature fusion based video fire detection algorithm.In this topic design. It based on machine vision of early fire warning research and design. In the image acquisition phase, To supervise the site monitoring is based on CCD camera, and to acquisite a picture, and the image from disk memory buffer is copied to the disk on the clipboard.In the image analysis stage,it can extract image of relevant information from the clipboard, it is based on a comprehensive consideration of the whole system, after image analysis, it will get a flame pixels in the whole image in proportion, and then to realize decision stage through the data design relevant decision criteria.Design of our subject:First of all, the VFW acquires lots of images, then it processes and analyzes the images collected to see whether the flame image area broaden or narrow. Finally, we perform color discrimination and elimination of non-fire colored moving objects, it will take early warning actions according to the setting criteria. So we can use it as a functional module in the field of security monitoring.On the one hand it expands the function of monitoring, on the other hand, it increases the effectiveness of early detection of fires which has practical significance.