Research On Path Recognition In Intelligent Vehicle's Vision Navigation

In order to alleviate pressure on the earth's energy resources, improve vehicle's safety, comfort and maneuverability, the research of intelligent vehicle(IV) has got extensive international attentions. As a product of comprehensive discipline, IV involve AI, computer vision, mechanical design, control science and vehicle engineering, etc. Relying on the visual information, IV vision navigation synthesize, analyze, understand the obtained external visual information, so it could guide the IV running in the right way. Nowadays, vision navigation has become one of the hot spots in the IV's research domain. The key skill in it--path recognition, plays a vital role in the performance of IV navigation.In this thesis, develop a path recognition system of IV's vision navigation based on FPGA(Field Programmable Gate Array).The main function of the system is processing the recognition of the dynamic path image, then determining the distance and angle between the IV's longitudinal axis and road sign line, finally sending the results to the upper computer, providing the basis for follow-up direction control.First of all, based on consulting sino-foreign relevant literature in detail, this thesis summarizes the research background, status and trends.Second, this thesis outlines the FPGA design methods, and proposes the overall design project of path recognition system based on FPGA, divides the whole system into six major modules, brings about the hardware and logic circuit design of each module.Third, researches digital image processing. According to the specific requirements of the text, process the image in accordance with image data acquisitionâ†'image pretreatment(Median-Filter)â†'optimal thresholding binarization processingâ†'binary image morphological "opening" operation processing(denoising, repair).Finally, constructs the road model in the ideal case, studies the positioning of the road sign line and the access of the angle in the model.The result shows:the path recognition system designed in this thesis is simple, practical, and real-time, provides some foundation for follow-up studies.