Research And Implementation Of Driver Fatigue Detection System Based On VMD642

With digitalization and intellectualization, fatigue driving detection is a key technology for driving intellectualization and safety. This thesis studies fatigue detection based on videos with VMD642-C development board from Magic vision Company and CCS3.1 development software as the development platform, and TMS320DM642 of TI as the core processing chip.In this thesis, decoder chip TVP5150, driven by HC trunk, captures the PAL format analog signals acquired by modified 850nm CCD infrared cameras, compiles them into YCrCb4:2:2 format digital video signals of BT656 and sends them to TMS320DM642 for image processing. After image preprocessing, adaptive threshold segmentation and binaryzation, signal frame images perform horizontal and vertical integration projection to define the face region and obtain the approximate eye region according to the redefined proportion segmentation of three courts and five eyes, and finally use Sobel operator to acquire the maximum angular point to define the eye position. Inter-frames distinguish fatigue with improved P80 standards based on frame numbers and control the LED on the development board for flashing alarm through CPLD. Then, one copy of the image demarcating the eye region is DA-encoded with saa7121 encode chip and TVP5150, translated into analog signals and sent to short video displays for display, and another copy adopts UDP transport protocol for network transmission so as to directly reflect the eye demarcation and fatigue test results on the far-end PCs. In this paper, the PCs are added with such functions as video capture and picture (in bmp form) storage to meet the needs for further study.Finally, it makes a quantitative analysis of the experimental results, and puts forward a work plan for the next step based on the software-hardware conditions for experiments and the known research method.