Method and device for traffic light image processing

Abstract

The invention discloses a traffic light image processing method and equipment, which are used to solve the problems of the low detection accuracy of the existing traffic light state recognition algorithm and the possible mispainting of the traffic light position. In the embodiment of the present invention, the first position area and the color state of each traffic light in the first rectangular frame are determined through the neural network model. If the color state of the traffic light is red, the first position corresponding to the traffic light detected by the neural network model A location area is compared with the second location area corresponding to when the color state of the traffic light in the previous N frames of traffic light images is red, and if the distance between the first location area and the second location area does not exceed the second preset threshold , the pixel to be processed is determined by the first position area; otherwise, the pixel to be processed is determined by the second position area, and the position of the pixel to be processed is determined by multi-dimensional information, so as to improve the accuracy of image processing and avoid a single neural network model Wrong location information output leads to image misprocessing.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method and device for image processing of traffic lights. Background technique [0002] At present, electric police cameras in traffic management are widely used to capture vehicle violations at intersections, providing a basis for punishment for traffic violations. It mainly generates a violation evidence map by recording the violation vehicle behavior and traffic light color status information, so that the punishment of vehicle violations becomes justified and well-founded, and punishment disputes are avoided. [0003] Under different light intensities, the color imaging of traffic lights in electric police cameras is not necessarily the intuitive red, green, and yellow colors. For example, in scenarios where the ambient light is insufficient or the ambient light is too strong, the center of the red light may appear yellowish or white, and the center of the green li...

AI Technical Summary

Problems solved by technology

[0006] The present invention provides a traffic light image processing method and equipment, which are used to solve the problem that the detection accuracy of the existing traffic light state recognition algorithm is low, and the color of the traffic light may be incorrectly painted.

Images