The Research Of Machine Vision Information Processing Algorithm In Smart Home System Home Control Center

As the product of the high-tech society times, Smart Home has been paid wide attention to by people and businesses in recent years, which represents a new life style of intelligence, convenience and safety for people yearning. However, at this stage of Smart Home mainly focus on the applications of electrical appliance intelligent control and access control security system, these applications of Smart Home do not really meet the needs of users and can not be widely popularized in the market. With the home frequent accidents of the elderly, children and the disabled people, the Home Furnishing safety accidents, people gradually shift the focus of Smart Home to these special populations. This paper puts forward the research of machine vision information processing algorithm in the Smart Home system home control center, both in the application or the promotion of Smart Home, has a very important significance.Firstly, this paper puts forward the overall design scheme of the intelligent home control center and peripheral modules and the design of intelligent home control center communication monitoring software, including μC/OS-II real-time operating system transplant, LwIP protocol stack transplant, the realization of network communication protocol and machine vision information processing interface design.Then, in three different application scenarios we research the algorithm of the home control center machine vision information processing. First, according to the Smart Home security, the algorithm of indoor environment anomaly detection is proposed. For the situation of stranges into the home, a new threshold is generated by the complementary of dynamic threshold and static environment compensation, which has solved the problem of the differential image binaryzation is easily influenced by illumination changes which can not accurately extract the suspected foreground region in the background subtraction method. For the abnormal situation of smoke, an improved running average method for background modeling is proposed, which overcome the the background update problem caused by the smog chronic expansion. Meantime, in order to achieve a more accurate prediction, puts forward the method of alarm factor segmentation threshold.Second, according to the safety problems of the elderly living alone, the algorithm of the human body posture detection is proposed. We use the background modeling based on the improved adaptive hybrid Gauss function to achieve the body gesture modeling and extract the feature vector. Then, identify the body posture through the hierarchical identification method to realize the judgment of the abnormal detection, so that issued a warning timely.Third, according to the home intelligent control, we study the algorithm of digital hand gesture recognition. Achieve the body gesture modeling through the introduction of the relevant algorithm of the median filtering, contour tracking, DP polygon approximation and the convex hull and convex defect feature. The SVM classifier is used to realize the simple digital hand gesture recognition so as to provide a basis for subsequent use gestures to control smart appliances.Finally, to design mobile phone terminal monitoring software based on Android system, realized data monitoring and control between the mobile phone APP and home control center in the Eclipse platform.