Technology Of Vehicle Speed Detection Basic On Video

Video traffic information collection technology is a hot research field today, speed detection is one of its key points. This paper's speed detection method is dividesd into four steps:projection, eigenvalue, look-up table, fitting. Mapping relationship is the base of look-up table and the key to speed detection with high accuracy. This paper proposes the following improved methods on the basis of past researching:1 Build two-dimensional mapping relationship, propose filling the empty pixel with the nearby value, this is a more simple approach of filling which is different from the original linear method, and comparing whth the original idea that looking for pixels from a distance propose a new idea finding the distance from the pixel, offer another two arithmetics which do not need to fill empty pixels to establish the two-dimensional mapping tables.2 Offer the dynamic average value method and the maximum value ratio method for the search of eigenvalue, on the basic of minimum value method, maximum value method, average value method and valley value method, which improve the adaptability of image processing and anti-jamming.3 According to the principle of least squares, regard the eigenvalue after look-up table as sampling value, excogitate the linear fitting arithmetic based on the polynomial fitting arithmetic, adding the second fitting, which improve the accuracy and efficiency of speed detection.A lot of experimental data obtained using the video speed detection proposed in this paper in some tunnel and Outer Ring Road of some city show that the arithmetics can timely detect vehicle speed with accuracy rate of more than 90%. And the results of small quantities sample application show that the arithmetics for speed detection can be used not only in tunnel and urban Outer Ring Road but also freeway.