Research And Implementation Of Low Illumination Video Imaging Based On Embeded Linux System

Monitoring equipment depend on flashlight when working on the highway in the night,it is against the international standard, and it will make drivers feel dizzy. So lowillumination camera must be the development direction in the future. Not only onemonitoring camera will be installed in the same scene if we want to the whole scene view,but the frames are dispersive in the monitoring terminal, it is harmful to analyze videoinformation, and it will cause the visual fatigue. So using image stitching to get a largemonitor scene is also the development of the future. This dissertation is based on theTi-Davinci-DM8168multicore hardware development board, to realize the image stitchingapplication program based on this hardware. Firstly, design the optical system and themechanical structure applied to this project. And, design an image stitching algorithm basedon matrix. Using C language and assemble language to complete the algorithm andtransplant to the development board. Achieve the video image acquisition, splicing,encoding, storage, and online preview display function.Dissertation’s main research is as follows:(1) Design the optical system and the mechanical structure applied to with this projectThe thesis researches and designs the optical system and the mechanical structure ofthe image acquisition system. Design a remote storage system, make it easier to machine,have a higher precision and reproducibility, ensure consistent of overlap region.(2)Design and implement an image stitching algorithm based on matrixIn order to meet the special requirement of this project, the real-time video imageprocessing, the mechanical structure and optical system are fixed and a lot of overlap region.Design and implement an image stitching algorithm based on matrix, this kind of algorithmis very efficiency.(3) Design and implement data link and transplant to development boardUsing software development kit (DVR-RDK) that provided by Texas InstrumentsIncorporated. Set up the data link using the API in the DVR-RDK. Achieve the video imageacquisition, analog-digital conversion, de-interlacing, de-noising, ARM (A8) controllingdata using ARM (A8), stitching image using DSP (C674x), video encoding and decoding using VICP (M3) and displaying the video.(4) Design and implement the way to promote efficiencyUtilizing optimized on code level, DMA data transmission, set the optimal compileroptions and linear assembly programming to improve image processing speed, meet the ofexpected processing requirements.