An Investigation of Fire Detection

• Main Authors: Jian-Wen Peng, 彭建文
• Other Authors: Wen-Bing Horng
• Format: Others
• Language: zh-TW
• Published: 2001
• Online Access: http://ndltd.ncl.edu.tw/handle/89544312207866452133

碩士 === 淡江大學 === 資訊工程學系 === 89 === In general, the object tracking technology can be applied extensively to traffic control, public security, and object recognition. About object tracking technology, there are three main steps of tracking the object from a series image: separating object from the background, identifying and tracing it.In terms of traditional object tracking technology, the tracked object usually should have fixed shape, so it can be separated from the background by characteristics. For instance, the tracked object can be any line-moving object, such as running cars, flying airplanes or sailing boats. The tracked object can be separated by the general technology due to the monotone of background. However, this kind of traditional technology fails to track the object that rapid moving or continuously changing in shape.The purpose of this study is to discuss a new technology called “Color Separated” that tracked object from the viewpoint of chromatology. It also can track the object that moves rapidly or keeps continuously changing in shape, especially natural phenomenon, such as fire or water. This study put more emphasis on the tracked object’s color of the collaboration between hue, saturation and brightness in images. This technology can track rapid-moving and unfixed object with the specific color bandwidth precisely. Moreover, the technology is not necessary to consider about the various problems, which are interfered with resolution, noise signals or different shot angles and focus.“Color Separate” technology can be applied to fire detection. In this study, a “fire-detect system” model is built based on “color separated” technology. Compared traditional methodology of fire-detect with new technology, the area of target image can be found by subtracting the object images and getting the range of image differences. In terms of “color mask” of “Color Separated”, the area of “fire” image will be separated from background image, then the system is able to distinguish the real color of “fire” image from similar color of non-fire object, then non-fire object will be discarded in order to identify the real “fire” image by way of “color mask”.In the model, it should be mentioned that “disorder” is used to be instead of traditional complex operation. Disorder can find fire rapidly and detect the range that the fire spread. In addition, it can also measure the burning area and burning level.In summary, this system is able to operate properly real-time. Combined with monitoring and alarm system, this model can provide proper warning to avoid the disaster as fire just starts. It is great helpful to industrial security and public security.