Double-mode intelligent vision sensing system architecture

Abstract

The invention discloses a double-mode intelligent vision sensing system architecture, which is characterized in that: the signal output / input end of a double-mode intelligent camera is correspondingly connected with the first signal input / output end of a broadband transmission network, the second signal input / output end of the broadband transmission network is correspondingly connected with the first signal output / input end of an application management server, and the second signal output / input end of the application management server is correspondingly connected with the signal input / output end of an application client. The architecture has the advantages that: an intelligent video processing server is saved; the task of the module is placed at the front end of the camera, so the advanced intelligent processing function is realized, the early equipment investment is saved, and the equipment maintenance cost and the system operating cost are reduced; real-time video analysis result-based image capture and automatic target tracking of real-time cradle head control are realized; parameters such as camera sensitization, illumination, contrast and the like of main body conditions serving as targets are automatically adjusted; and the imaging quality is higher, and the analysis result is more accurate.

Description

technical field [0001] The invention relates to a dual-mode intelligent visual sensing system architecture, which can be widely used in technical fields such as intelligent transportation, intelligent security, video conference, medical expert system, and automatic production quality inspection. Background technique [0002] The structure principle of the existing system is as follows figure 1 shown. The existing intelligent visual information is collected by the network camera and compressed and coded. It is connected to the video processing server through the broadband transmission network, and the video decoding, pattern recognition, artificial intelligence and other algorithms are run on the visual server. Intelligent vision servers are usually high-end server computers or industrial computers. The vision server transmits the analysis results to the application management server. The application management server serves each client with the result of intelligent video...

AI Technical Summary

Problems solved by technology

[0003] 1) Real-time detection cannot be realized

Due to the mechanism of video network transmission, there is a delay of several seconds between the video reaching the intelligent processing server and the image collected by the front-end camera, which has an impact on illegal pursuits, etc.

2) Unable to automatically adjust the camera in real time

The delay of several seconds makes it impossible to automatically adjust the camera acquisition function (such as lighting exposure, automatic PTZ control target tracking, target zoom, etc.) through the results of intelligent analysis.

3) The pressure on transmission bandwidth is very high

For multi-channel high-definition video surveillance, each channel requires 5M bandwidth, which brings great pressure on transmission equipment and system operating costs

4) The image definition is not high enough

Such a resolution cannot meet the requirements of multi-lane vehicle detection algorithms

5) High requirements for the video server in the background

It is uneconomical to equip a video server at the front end of each camera, and the video server requires a lot of calculations to process the decoding of compressed video, and it requires an expensive high-performance computer to process the video streams of multiple cameras at the same time

6) If a video server handles multiple front-ends, it is easy to cause problems such as loss of key events and unreliable systems due to running a non-real-time operating system

7) The system is complex and the maintenance cost is high

A server failure can cause multiple camera fronts to become blind spots

Images