Silhouette-based multi-sensor smoke detection

摘要: Fire is one of the leading hazards affecting everyday life around world. The sooner fire detected, better chances are for survival. Today’s alarm systems, such as video-based smoke detectors, however, still pose many problems. In order to accomplish more accurate detection and reduce false alarms, this paper proposes a multi-sensor detector which takes advantage different kinds information represented by visual thermal imaging sensors. analyzes silhouette coverage moving objects in long-wave infrared registered (~aligned) images. registration performed using contour mapping algorithm detects rotation, scale translation between multi-spectral geometric parameters found at stage then further used coarsely map images them calculated. Since invisible its will, contrarily ordinary objects, only be detected As such, silhouettes will start decrease case smoke. Due dynamic character smoke, also show high degree disorder. By focusing on both behaviors, system able accurately detect Experiments non-smoke sequences indicate that automated Furthermore, low-cost analysis, fast warning, with low number can given.