A study on the flame detection and object classification technique using the color information

With the development of construction techniques, the high-rise buildings have been established and populated. In the event of fire in these buildings, as the fire is spreading, the risk of large fire is increased and the resulting casualties and property damages are increased. Therefore, the techniques to detect the flame at an early stage are necessary in order to prevent the fire and minimize the damage. The flame detection technique based on physical sensor has limited disadvantages in detecting the fire early. In addition, this technique has a problem of increasing the cost in proportion to the number of regions where it is installed. Image-based flame detection technique has been studied to complement these problems. In this paper, the object is detected by using the adaptive background modeling technique. In the flame detection technique, the flame is detected by converting RGB images to HSI images which have strength even in the change of lightings or scenes. Also, if the object with similar color to the flame is detected from the output image, it will detect it as a fire, leading to increase the misdetection problem. To solve these problems, objects are classified by using the direction information of the object detected. Based on experiment results, if the flame is detected only using the color information, it is confirmed that the fire is detected in real time. However, the problem of classifying the object as the fire occurred when the object had the similar color to the flame in the image. For this, it is confirmed that the performance was improved when classifying the object using the direction information of the object.