Research On Video Capture And Video Analysis Technology In Smart Home

We can get the infomation of our home by viewing the collected video images of our family house. With the video images can be saved and transmitted, so we may view our house anytime and anywhere. What’s more, as the video analysis technology developed, which can be applied to family video images for offering people decision-making information by analyzing family video. Moreover, video analysis technology combined with network and automation technology, intelligent operation can be made based on video analysis results.In consideration of that home video capture devices have different requirements, as low cost, low power consumption, easy to assemble, easy to expanded, good stability et al., this paper used embedded technology to colloct video. The proposed video capture system combined ARM processor, Linux embedded system and USB camera. The captured video was compressed using advanced H.264compression coding standard, then the compressed video was tansmitted to home PC via network cable. In addition, video can be uploaded to the internet, and remote PC, smart phone et al. can access to the video.Proposed system does the video analysis work on PC. First, in order to make people obtain meaningful information fast, this paper applies the shot boundary detection technology to home video, and the technology split the video into a series of meaningful shots. For observers, shot boundaries are produced when video content changes, then we can easily get the changing images of video by shot boundaries. In this paper, shot boundary is detected using Biased-SVM in H.264compression domain. Boundary features are got in H.264code stream, and shot boundary frames are classified by SVM. Second, to saving memory space, redundant home video frames are detected in H.264domain and removed to reduce memory space. Finally, vision-based hand gesture recognition technology applied, this paper designed a hand gesture command controlling system for smart home, defined a set of concise hand gesture commands, and carried out the recognition of these commands. The recognition results can be used to control home electrical equipments, and it makes the controlling system more intelligent.Experimental results indicate that the video capture system works stably and has some application value. The proposed shot boundary detection meathod based on H.264domain has a better detection performance, which divided the home video into shots, and we can view meaningful information quikly. The meathod of video storage optimization can effectively reduce memory space without breaking meaningful information. The designed hand gesture command controlling system adds the smart home interaction mode, the commands are concise and the controlling mode is nature.