Research On The Technologies Of Video Smoke Detection Based On FPGA

In a fire, smoke appeared earlier than open flame, so fire could be alarmed superearly through detecting smoke. Video smoke detection systembased on digital imageprocessing technologyhad better accuracy and real-time performance than traditionaldetection methods based on sensor. Nowadays, most video smoke detection systemwere designed by using computer, those systems were not good real time whenprocessing large amount of dataas computer executes instructionsserially. FPGA hadabundant internal resource and it could process dataconcurrently. The real time propertyof those image processing systems with large amount of data could be effectivelyimproved by using FPGA.On the basis of digital image processing, this paper research on a Video smokedetection system based on FPGA.Through a comparative experiment, median filter waschosen to denoise the original image. Considering about the difficulty and actual effectof foreground extraction of some different methods based on the relation of frames,background subtraction based on Kalman filter was finally chosen for foregroundextraction and backgroundupdating. In this paper, the connected domain of foregroundwas obtained by morphology processing the original binary image. Then BayesianDecision Theory was used to explain the necessity of using multi-feature in smokedetection. Analyse about the feasibility of using some features such as change of area,circularity, high frequency energy decay, constant gray level of smoke in smokedetection and consider how to extractthose features.On this basis, an algorithm fusedthose4features mentioned before was proposed for smoke detection and its actualresult was compared with the algorithm based on only one feature by experiment. At theend of the paper, thedata processing part of system was divided into several independentmodules. During the processof translating these modules into Verilog HDL some skillsof tacklingdata by FPGA were taken into account, simulation was then provided toreflect the feasibility of building a Video smoke detection system based on FPGA.