Research On Driving Fatigue Detection System Based On ARM Platform

It is known that fatigue driving has become one of the important reasons for traffic accidents. The traffic safety problems it causes have gradually raised widespread public concern. Therefore, preventing fatigue driving has become one of the main hot spots of current research.This article focuses on an proposed and implemented algorithm, which is improved based on ASM method for face and eye location:Because the results of position is not ideal in rich expressive, gesture changeable and non-uniform illumination conditions, the paper proposes a 2D cross-shaped Gabor model, and then combines the improved model and Gabor feature; It means to establish a 2D cross-shaped search area in the center of the feature points. As probability density estimation model is established in the search area, judging the optimal search position by the probability distribution difference model is accessible; The algorithm is improved by combining the improved local feature and Gabor wavelet feature. The method increases the search area from linear area to 2D cross-shaped area. Feature points’gray information is increased and the effect of light can be filtered out. That is why the accuracy and robustness of the algorithm are improved.To further improve the accuracy of eye positioning, this article proposes integral projection algorithm combined with local ASM way with the aim to achieve twice positioning the eye.Then the final position of eye is determined according to principle of minimum ahd weight parameters.Based on this, the paper designs fatigue driving detection system on the basis of an ARM. At first, face and eye can be located by the improved ASM algorithm. Then the face area will go through integral projection process to locate eye area for the second time. The final eye area by weight parameters is determined after that. Finally, the use of state recognition method is based on the eye ellipse fitting, then on the basis of the discrimination method PERCLOS fatigue, combined with blink frequency identification method, and using these two parameters to determine the fatigue state of the driver.Hardware platform using a friendly arm’s embedded Tiny6410 platform and software platform has been built, including embedded operating system of choice, the establishment of cross-compiler environment, structures, the overall design of the system software Qt/E of Transplantation and integrated development environment.Finally, the test system based on ARM is implemented. And the experimental results show that the total correct rate of the system is 91.31%, which indicates that the system can meet the needs of driver fatigue driving detection, and have some reference value.