Research On Vehicle Recognition And Distance Measure Based On Span Projection Histogram And Corner Point Detection

Malignant transport event in highway is a serious social question in the era when highway increases rapidly, which not only leads to family's economic loss, but also menaces people'lives. So the demand of studying vehicles'detection and distance is urgent. In the paper we have studied AOI abstract algorithm and vehicle and distance measure algorithm. At last we have gotten the results as follows:(1)Classification and analysis of AOI abstract algorithm and vehicle recognition and distance measure algorithm.(2)Construction of structural road AOI abstract algorithm based on fan projection histogram suitable for both straight and curve lane with a certain curvature, resistant to stronger disturbance against AOI abstract algorithm based on model with high time complexity and low suitability for curve lane.Our method can be used to detect straight lane and curve lane with a certain curvature, which needn't strict driving direction and can adapt both to shade disturbance caused by vehicle shelter, more intense sunlight, and more serious disturbance caused by serious lane loss, long distance dashed lane and deceleration belt. It has adaptive ability to intense and weak light to some extent and stronger ability resistant to disturbance.(3)To solve the question hard to abstract feature and to locate goal, on the basis of robust AOI abstraction algorithm based on fan projection histogram, we further study vehicle detection and distance measure algorithm based on corner detection, which can abstract feature more easily and implement simultaneity of detection and distance measure. Experimental result demonstrates our method has a larger promotion of feasibility and robustness in vehicle detection and distance measure.Totally, It is studied deeply AOI abstraction and vehicle detection and distance measure on highway. Both a kind of AOI abstraction algorithm with more powerful adaptability and stronger robustness and a kind of vehicle detection and distance measure algorithm based on corner feature are presented. Experiment demonstrates efficiency and robustness of our algorithm in vehicle detection and distance measure.