An Automatic System for Smoke Detection in Outdoor Areas

Early detection of fires plays a crucial role in minimizing their impact and preventing them from spreading. Every year, the repetition of fires results in the loss of human life, animal life, and plant life. Fire detection has become increasingly desirable and significant in surveillance systems, where traditional methods of detecting smoke relied on smoke sensors. Therefore, this method is ineffective in open and large buildings, and outdoor areas. As a result, this study suggests using computer vision systems to detect smoke in open spaces by using a static camera. To reduce the data size while preserving important details, the input video is framed and decomposed using the Integer Haar Lifting Wavelet Transform (IHLWT). Then, for smoke color detection, a new method called the multi-threshold International Commission on Illumina (CIE) Lab color space is used, which took into account the smoke colors' change from whitish gray to blackish gray. In addition, the Frame Differences (FD) technique is used to detect motion and thus reduce false alarms. The smoke color detection is combined with frame difference techniques. The small pixels are removed via a morphological operation that