A Smoke Recognition Method Combined Dynamic Characteristics and Color Characteristics of Large Displacement Area

The current video detection methods about smoke dynamic characteristics is not comprehensive. And for backgrounds with insignificant texture features, extraction methods of suspect smoke areas only using color feature is not flexible enough for environmental adaptability, which leads to many missed and false alarms. To address the above problems, the color characteristics of images in RGB-HIS model and an improved fast Horn-Schunck optical flow algorithm [1], [2] are used to obtain optical flow field by which the image segmentation is performed on both suspected smoke motion area and non-smoke area and extract the large displacement area. Then, the mean direction and the average velocity of the optical flow vector in the motion area are calculated. Finally, the smoke and other interference sources are distinguished to determine the existence of the smoke. Results have shown that this method can effectively eliminate the common interference sources and identify the smoke area with high accuracy and robustness.