Font Size: a A A

Study And Implement Of Embedded Video System's Key Technology Based On Davinci Platform

Posted on:2012-03-20Degree:MasterType:Thesis
Country:ChinaCandidate:D Y RenFull Text:PDF
GTID:2178330335959869Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
Embedded video surveillance system is that integrates multimedia information technology, computer interactive technology,distributed communication technology and Real-time monitoring technology. With the development of network technology, multimedia technology and integrated circuit technology,Embedded video surveillance system is to digitize, internetize, intelligentization and integration.The thesis studied embedded video surveillance system based on dual-core processor DM6446.Firstly, DaVinci technology platform and software and hardware structure of dual-core processor DM6446 are analysed. Then two main algorithms are studied.one is automatic exposure(AE) algorithm and automatic white-balance(AWB) algorithm aiming at improving images'quality. Another is people counting algorithm based on moving object detection. At last, based on TI's dual-core processor TMS320DM6446, intelligent video system is designed and developed. Core modules of the system conclude video acquisition and preprocessing, AE and AWB algorithm, people counting algorithm video display module and alarm module.Common existing automatic exposure algorithms were analysed in the first. Taking both algorithm's efficiency and system's efficiency into consideration,a method of uniform partition with weight was proposed to calculate current image's average brightness and a gradual approximation means with different adjusting step-size was put forward to control shutter and gain.Automatic white-balance algorithm used in this thesis was based on grey world assumption.Then, the thesis analyzed moving object detection algorithms and studied image denoising method, background modeling method and post-processing of moving objects.On the basis of theoretical analysis,a people counting algorithm based on moving object detection was proposed.After image denoising using filtering technique,A adaptive background modeling method was proposed.With edge detection technology and morphological methods,people counting algorithm was implemented based on object size.Using TI's dual-core processor TMS320DM6446 as the main processor, a intelligent video system was designed.In this video system,algorithms were realized based on Codec Engine.And, this thesis used multi-thread technology to design and develop video acquisition and preprocessing module, AE and AWB algorithm, people counting algorithm,video display module, alarm module.At last,through system integration and testing,Correctness and feasibility of AEWB and people counting algorithms were demonstrated in the DaVinci platform. The experimental results showed that AEWB algorithm could improve quality of images very well in the platform. People counting algorithm could also achieve good statistical results.
Keywords/Search Tags:DaVinci technology, AE and AWB algorithm, moving object detection, people counting, embedded video system
PDF Full Text Request
Related items