Sequence Of Images Of Traffic Signs Detection Technology Research

Traffic signs automatic detection system is an important part of modern intelligent traffic management, it has a great research value. On the basis of color information and shape character this thesis has done the following research work on the aspect of locating traffic signs.First, this thesis puts forward a traffic signs positioning method which integrating of multi-channel. First this method divides saturation components of HIS channel into segments through threshold, roughly extract the region of traffic signs and inhibit the interference of the level of illumination; Then it solves the information missing problem caused by the division of the threshold by making use of the relevance of three color components in RGB channel. In order to improve the efficiency of positioning signs, this method making use of the relevance between the former image and the latter one in image sequences predestinate the position of traffic signs on the basis of the first frame being positioned. Experiments show that the integration of multi-channel approach can locate the position of traffic signs quickly and effectively from the video image sequences.Secondly, by analyzing the characteristic of traffic signs shape, this thesis puts forward a method that judging the shape of traffic signs by using the least square method which is based on curve fitting after making a projection in the horizontal and vertical direction. Experiments shows that using this method can not only get the shape of traffic signs, internal designs and character information, but also reduce noise interference effectively.At last, according to the fact entropy of information has the characteristics of translation, zoom and rotating; the thesis promotes an evaluation function to detect confidence level which is based on entropy vector. What's more, the thesis also modifies this criterion by combining the distribution regularities of the traffic signs in the image sequence. The result shows that using this method can effectively exclude some misleading objects which are similar to the traffic signs on color and characteristic of shape.