Design And Research Of Video Emergency Communication Software Based On Easydarwin Platform | | Posted on:2019-05-01 | Degree:Master | Type:Thesis | | Country:China | Candidate:H You | Full Text:PDF | | GTID:2438330548458406 | Subject:Engineering | | Abstract/Summary: | PDF Full Text Request | | The purpose of building an emergency communication system is to play a temporary function when the original communication system is destroyed.In many large natural disasters and public emergencies,conventional communication methods are often destroyed.At this time,the construction of emergency communication system is very important.In certain circumstances,the cost of bandwidth must be taken into account in the establishment of an emergency communication system.In view of the limited resources after the accident,the main way of communication is satellite communication(limited bandwidth),the communication equipment is a low performance terminal device(performance is limited),and there is not enough manpower to monitor the scene conditions(human resources are limited).In most video emergency communication systems,in order to accurately grasp the scene of the scene.We often use high quality and clear picture video sources and high fidelity voice input,and send multiple people to monitor the scene in real time.This paper proposes a solution to the three point resource constraints mentioned above.In the case of limited bandwidth,we proposed a method decode and encode the initial HD video stream and to reduce the occupied bandwidth.Then,we proposed a way to reduce the bandwidth by using text information instead of the original voice information.The second research work of this paper are designed to reduce the computing work of the terminal equipments,we only extract the acoustic features of speech utterances and directly send them the server for decoding instead of using the raw speech files.The third research work of this paper is aimed to reduce the use of manpower.Fire detection is carried out in the remote emergency command center,and it can accurately identify whether there is a fire on the unattended. | | Keywords/Search Tags: | Easydarwin, FFmpeg, RTSP, H264, Tensorflow, object detection, kaldi, speech recognition, UDP | PDF Full Text Request | Related items |
| |
|