Abstract:During the drying process by microwave, the image of illuminant of fire and flashlight will become blur due to the?motorial material and the differences of color, size and the degree of reflection when material gets fire, which make it difficult to extract features. This phenomenon makes the images of transmission have lots of noise which contributes to the high rate of wrong alert by temperature detector and infrared detector and even leads to the machines out of service. Aiming at the particularity and complexity of microwave equipment to detect fire, we build a fire warning model for microwave equipment based on the research that the image of fire is collected by CCD with protective device in order to solve the problem that traditional instruments sometimes are out of service in complex working environment. During the image processing, we combine the recognition of flame image color and the recognition of gray level brightness in case of wrong alert. We target to remove noise on images by adopting conditional statement cyclic structure algorithm and embedded threshold filter program as a complementary. The USB ports are connected to hardware interface of NI myRIO. With the help of LabVIEW software programming embedded in hardware system, we process the received image. The result of analyzing the datum shows that this system can overcome the problem of wrong alert by adopting temperature，smoke，?photon，gas，composite detectors.
