A statistical method for the color cycle of traffic lights

Abstract

The invention provides a method for counting a traffic light color period. The method comprises the following steps: A, detecting a video image of a current traffic light color; B, judging whether the detected color is accordant with a currently-determined traffic light color x or not, if so, updating the count value iColcntx of the current traffic light color x, and returning to the step A; otherwise, skipping to a step C; C, updating the count value iColcnty of a detected result color, and judging whether iColcnty is larger than a preset traffic light color switching condition value Ny or not, if not, returning to the step A; otherwise, turning to a step D; D, confirming that the current traffic light color is y, updating the counting times CycleNx of the traffic light color x; E, judging a current confidence coefficient level according to the value of CycleNx, and updating the period counting result of the traffic light color x by using formulae which correspond to different confidence coefficient levels. According to the method, the detection result of a traffic light video image is counted periodically, and a corresponding traffic light detection period model and a correction mechanism are established, so that the functional accuracy and usability of electric warning equipment are enhanced.

Description

technical field [0001] The invention relates to the technical field of video monitoring, in particular to a method for counting the color cycles of traffic lights. Background technique [0002] At present, in the pure video detection method of judging traffic lights in the electronic police equipment for running red lights, the input YUV image is mainly processed by median filtering, binarization, pattern learning and other methods to obtain the brightness information and chromaticity information of the input traffic light image. . After complex logical confidence judgments, the test results are obtained. However, in actual scenes, due to the influence of light, weather, camera function, and other physical factors, this detection method is prone to different degrees of detection errors when dealing with some special scenes. For example, in the forward and reverse light environment, the brightness and chromaticity information is invalid due to the sunlight shining on the li...

AI Technical Summary

Problems solved by technology

However, this method has the following problems: it is not easy to determine the special lighting time as it changes with the seasons; the traffic light template in special scenes is complicated to make and lacks universality; it cannot solve physical phenomena such as traffic lights shifting on the screen.

Images