Font Size: a A A

Design Of Indoor Panoramic Monitoring System Based On DaVinci Platform

Posted on:2017-01-13Degree:MasterType:Thesis
Country:ChinaCandidate:W Z DuFull Text:PDF
GTID:2308330485492779Subject:Embedded image processing
Abstract/Summary:PDF Full Text Request
Video surveillance is one of the most basic facilities security systems and the comprehensive-ness is the cornerstone of the surveillance. However, the traditional designs on ultra-wide area surveillance in real-time and low price use still have some restrictions. In this thesis, we proposed a single rotate camera to capture image and get panoramic image in real-time with an surveillance system based on TMS320DM6446, which is one of Ti Da Vinci series. Based on the relationship between stitching matrix and the location of image captured, we built the soft sensor model and re-polished it to be suitable. It showed that the system operated stably and has great practical value.This thesis elaborates the inner panoramic surveillance system based on DaVinci platform, the contribution and innovation are as followed:1. Set up a hardware and software system on DaVinci platform. The core for this system is DM6446, a dual-core control chip with both ARM and DSP. The ARM realizes the image acqui-sition, management, display, network communications, data management and other functions with MontaVista Linux real-time operating system. The DSP handles the image stitching algorithm with DSP/BIOS high-performance operating system. Dual-core work together efficiently.2. Build a mosaic matrix model under the rotating camera. The phase correlation method is utilized to calculate the image position, while the SIFT operator is applied to calculate the mosaic matrix. Therefore, the mosaic matrix is obtained by SVR model based on the relation between the image position and mosaic matrix. Then the panoramic image with phase correlation is polished to get the available panoramic image for monitoring.3. Update server-based model. The core collect images of clockwise and counterclockwise to server periodically. The server then downloads the updated SVR model parameters based on image sequence. After that, the model is transferred to the core. The model updating process guarantees that the monitoring performance is not affected by the environment change and system aging.4. Provide remote monitoring capabilities. An APP is designed with user login authentication, supporting the current and historical indoor panoramic image search. Also, an interface with indoor environmental parameters is provided to make the system more user-friendly.
Keywords/Search Tags:DaVinci, video surveillance, image mosaic, phase correlation, SVR, SIFT
PDF Full Text Request
Related items