Indoor activity detection and identification method and system based on multi-sensor network

Abstract

The invention relates to an indoor activity detection and identification method and a system based on multi-sensor network wherein the method comprises: recording the position information and triggering information of all the triggering type sensors in a targeted indoor environment; prior to the identification of the activity type, training and processing the k-nearest neighbor algorithm to obtain identification bases; in the practical application, according to the processing of the triggering information, the heterogeneous pheromone residual mask and the single frame picture change to obtain a corresponding a pheromone matrix; and finally, based on the identification bases, realizing the identification of the human body activity type. According to the invention, a plurality of triggering type sensors are arranged in an indoors environment and the triggering information and the position information of each triggering type sensor are recorded, which increases the accuracy and the reliability of the identification result considerably.

Description

technical field [0001] The invention relates to the field of environmental assisted living, in particular to a multi-sensor network-based indoor activity detection and recognition method and system. Background technique [0002] In recent years, in the indoor environment, people have higher and higher requirements for location-based service quality, in order to detect the real-time location of human activities and identify the types of human activities, so as to drive intelligent decision-making and provide human beings with the services they need. Or turn off indoor facilities such as lights and air conditioners. Therefore, in order to improve the service to human indoors, it is necessary to detect human activities. Contact sensors and non-contact sensors are now widely used in indoor environments, and there have been many successful explorations in smart buildings and smart home environments based on multi-sensor networks. [0003] In the home environment, user activity ...

AI Technical Summary

Problems solved by technology

[0004] However, for the activity recognition method based on the trigger sensor network, the existing technical solution is generally an analysis solution based directly on the trigger sequence, mainly by directly triggering a certain sensor or triggering certain sensors in a specific order as related activities. This method ignores the location information of the sensor, and the judgment basis is relatively single, which is prone to misjudgment, and the data readability is poor, and only the final activity recognition result can be obtained.

For the activity recognition method based on indoor positioning system, because indoor areas such as dining table, sofa, bed, etc. can correspond to various activities, these areas are called activity ambiguous areas, so the type of activity of the target cannot be accurately judged only by positioning information

Therefore, the current identification methods for user activities cannot accurately identify and judge the type of target activity, and the reliability of the target activity identification results is low.

Images