Method And Implementation Of The Smoke Detection Based On Image Processing

In recent years, the development of image processing technology is rapid, the fire detection based on image processing technology is one of the important methods for fire monitoring. The smoke is often the most important characteristic of the fire before open flame. Compared with traditional detection methods, video smoke detection image processing can greatly reduce the probability of false alarm rate and missing alarm rate. At present, domestic research is almost the first image acquisition and recognition processing based on PC, and it is not timely. We make full use of FPGA which can achieve data parallel processing, we put the video image processing method used in the FPGA platform, it can realize real-time fire detection well.Smoke detection method is proposed in this paper and the realization on FPGA platform based on the image processing. We use the high sensitivity and low voltage OV7670 CMOS VGA camera to acquire the images. We use the improved median filtering method to remove the noise of original image. We analyze several methods of background updating and foreground extraction. Improved surendera background updating algorithm is used, then do two binarization and morphology processing. We extract irregular edge of smoke, flicker frequency characteristic of its edge, variation characteristics of internal pixel, diffusion characteristics and consistent with the characteristics of gray, and so on. Finally, we put these features by feature fusion method to judge, and quantify these features and fusion method, finally we realize by dividing the modules parallel processing method on the FPGA platform. It proves the feasibility of video smoke detection method in image processing based on FPGA platform by Simulation and practical results.