Research And Implementation Of The Driveway Security Warning Algorithm Based On Machine Vision

With the rapid technological development trend, the driveway safety based on Machine Vision is an important part of the modern intelligent driver assistance systems.This Driveway Security Warning System focuses on the research of lane departure detection and barrier detection two aspects.The algorithm can detect the white lane, the yellow lane and the barrier or abnormal on the front road. The paper use LCD display and voice playback "pay attention" two ways to early warn.Firstly, the paper according to the collection vision angle of CCD and width of the driveway to determine the lane departure detection and obstacle detection target area; According to the YUV format image features, the system use digital image processing technologies to get gray image, use the improved fast parallel median filtering and gray stretching to enhance image features, use improved45°and135°Sobel diagonal operators to detect edges of the image, use two-dimensional Otsu adaptive threshold algorithm to get binary image; Then the system use improved pixel marking method to label the binary image connected domains, the target area is divided into two regions to statistic the largest connected domain respectively which is used to fit the left and right lane; Finally, the system use the least square method to calculate the slopes of the left and right driveway lane, use the slope to calculate the angle as the standard to judge the vehicle whether or not deviated from the lane. In addition, according to the characteristics of the connected domains of the middle region of the target image, such as the number of pixels and the horizontal distance of the largest connected domain, the system test the barrier or abnormal of the front road, combined with the space between the actual vehicles’ left and right tire to determine whether to warn.Software simulation part of this system, this system use MATLAB to simulate the above algorithms, comparisioning and analying the commonly used algorithms to select the most suitable algorithm. Such as the testing of the improved median filter, the sobel edge detection and two-dimensional Otsu adaptive threshold algorithm which demonstrate the feasibility and superiority of the algorithms.Hardware realization part of this system, we use ICETEK-DM6437-B-KIT of the Beijing REALTIME to design a DSP hardware development platform. DM4637is the core processor of the platform, a TFT LCD screen and a day and night CCD camera are also included; the stereo equipment for playing the voice alarm information. The system can achieve a real-time visual image acquisition, processing and results image display processing, voice alarm and so on.The experimental results show that the software algorithm has robustness and the hardware platform’s resource is plentiful and function fully. The system can achieve a good effect of lane departure detectin, obstacle detection and early warning function.