基于多光谱分离的烟雾检测

Abstract

Because of the limitation of lower anti-interference ability,traditional smoke detection technologies are hard to distinguish the fire smoke,water fog,grass and other false information.Therefore,a smoke detection method based on multispectral separation was proposed according to the Principal Component Analysis(PCA)and transform domain method.First,band pass filters in the regions of 460-520nm,540-570nm,and 580-610nm were used to capture the principal components in the spectra emitted by an objective under monitoring.Then,the plane of transform domain was build by space fitting for these principal components.With simple algebraic judgment on the plane,the classification and recognition of the smoke,fog and grass were completed.Finally,Monte Carlo simulation was performed for 1 000times.The results show that the three-band spectral analysis method proposed is simple,easy to operate,and the separability of P3and P4is 0.113.The method not only overcomes the shortcomings that cannot distinguish the smoke and water fog in traditional image and video detection methods,but also avoids complex spectral classification algorithms.It shows excellent practical values in the field of fire monitoring.