Design And Implementation Of Driver Fatigue Detection System Based On DSP Platform

Drowsiness driving has caused hundreds of thousands of deaths and disability every year, and absorbs attentions from various fields. As the research on fatigue detection technology in full swing, all these approaches can be divided into five categories according to its metric measures including subjective report measures; driver biological measures; driver physical measures; driving performance measures; and hybrid measures. The driver fatigue detection which we implemented it on the embedded development platform belongs to biological measure. And it uses Haar feature based Adaboost face detection algorithm to detect human faces, and uses PERCLOS criteria, eyes closure duration, blink frequency used as inputs to fuzzy inference system for fatigue testing.The driver fatigue detection system is adapted to different light condition and also functional while driver puts on sunglass when implemented in the embedded development platform. During the transplantation of the driver fatigue detection system, we first assess and analysis the required sources according to the processing sequence of the system. And then planning and designing the hardware resources based on the requirements. Realizing and implement the transplantation one by one based on the design, which divided the whole system into light judgment module, image acquisition module, voice alarm module, display module and storage module. After the transplantation, we refined the frequent used loop structures and computational algorithms. And improve the use of hardware according to the process ability of ADSP-BF533 chip. Finally, we present the system achievements, and the test of its functional range and the analysis of the errors.The main reason is the input image has missing information that causes driver fatigue detection system dysfunctional, which will be solve in the future.