Urban retina system based on mobile side end node

Abstract

The invention relates to a multi-level edge node camera shooting analysis platform for simulating an optic neural network, which is used for image analysis and ubiquitous monitoring in a city range. According to the system, a main camera, an edge computer and a communication assembly are arranged through a street lamp network, street lamps are changed into neuron nodes, and a visual private network covering the city area is constructed. Then, sub-cameras and edge computing power are arranged on pedestrians, vehicles and unmanned aerial vehicles, and movable edge end sensing nodes are formed. According to the method, an expandable city retina system is established, a point-line-plane combined visual network is formed together, comprehensive perception and beforehand prediction capabilities are provided, multi-level video analysis requirements can be quickly responded, and the intelligent perception requirements of a smart city are met.

Description

(1) Technical field [0001] The invention is a multi-level edge node camera analysis platform simulating an optic nerve network, which is used for image analysis and ubiquitous monitoring within a city area. The system uses the street lamp network to set up main cameras, edge computers and communication components, turning street lamps into neuron nodes, and building a visual network covering the city. Then set up sub-cameras and edge computing power on pedestrians, vehicles, and drones to form mobile edge-end sensing nodes. Further, the edge perception is integrated into the visual private network composed of neurons. The neurons are connected to the city brain through mobile phone signals, and a retinal system based on multi-level edge computing is established to complete the image classification and recognition system architecture and realize multi-dimensional edge perception. , fully explore the potential of ubiquitous sensors, and lay a technical foundation for the visual...

AI Technical Summary

Problems solved by technology

This model has been facing challenges such as large amount of data, difficult retrieval and difficult calculation, which can be summarized into three problems: difficult to store, difficult to retrieve, and difficult to identify

[0003] The first is that the storage cost is high, and the longer the storage, the higher the cost; secondly, there is no real-time feedback, and the server proposes the data for processing, which is poor in real-time and accuracy; thirdly, it cannot be collaboratively identified, and the information recorded by the A camera may not be available in real-time The B camera can be matched, but the recognition accuracy is very low

This "1-1 model" is increasingly unsuitable for the management requirements of smart cities

Images