Multi-channel H264 Video Smoke Detecting Algorithm Running On DM8168

In recent years, the automatic and intelligent detection and recognition system based on digital image processing has been well studied. The technologies of network video surveillance and digital image processing are the basement of H264 video fire detecting system based on DM8168 Soc.Ti's TMS320DM8168 Soc has an outstandingvideo processing ability, and has been widely used in video surveillance and intelligent video detecting system.This article presents a Muti-channel network video fire detecting system based on DM8168 Soc.And we have realized the video smoke detecting algorithm on DM8168 which is the core of the fire detecting system. The video smoke detecting algorithm supports multi-channel H264 video input,and has a good ability of detecting the smoke from the background in the early stage of fire accidents.In the future,more complicated algorithms will be applied to the DM8168, such as fire detecting algorithm based on two cameras, fire locating algorithm.This article has dived into the principle of motion detecting algorithm based on video frame difference with background and the video frame difference of two adjective frames.The article proposes the background updating method based on feedback model. And system filters the smoke-like disruptors according to the feature of frame wavelet energy decaying gradually and the feature of SSIM decaying gradually.The video somke detecting algorithm runs on the Ti DM8168 evm platform, and we develop the software to test the algorithm based on the McFW. This article has studied the mechanism of Links and Chains of McFW, and the universal method of XDAIS algorithm development.Here we make the fire detect agorithm running on the C674 x DSP core, and establish our own frame processing chain to test it. According to the result, the video smoke detecting algorithm has achieved the goals of Multi-channel H264 video smoke detecting.The result shows that this algorithm can detect the smoke well, and distinguishs the smoke-like distractors and background disturbance.