Image Sensor Based Fire Detection using Raspberry Pi

Every year the fire disaster always causes a lot of casualties and property damage. Early warning systems can significantly reduce the damage caused by fire. In this paper we present the state of the art of fire detection techniques using image processing. Furthermore, we analyse different types of models involve in detection techniques such as RGB and HSV models for detecting fire. The proposed technique utilizes prominent features such as flame colour information. The initial stage of the work extracts fire coloured pixels using a set of enhanced rules on RGB. Image processing through vision-based or sensor-based methods can be generally used to reduce the false alarm rate. Here we use a flame sensor to detect fire and the corresponding Pi camera image for image processing. The system is able to send notification to user precisely. The proposed system is successful with features such as easy installation, quick alert creation and simple usage in a closed room such as a store room or ware house etc.