A method for intelligently identifying the state of circular traffic lights

Abstract

The invention discloses a method for intelligently identifying the state of a circular traffic signal light. Firstly, the road image sequence is preprocessed, and the emphasis is on color component filtering. Then use the appropriate threshold in the HSV color space acquired by dictionary learning to perform color segmentation, and obtain three binary images, omitting the conventional grayscale image processing operations. Second, a dynamic multi-stage filter is designed based on the characteristics of circular traffic lights. Perform dynamic filtering operation to quickly screen out the candidate connected areas of traffic lights. Finally, the black body growth masking method is used to calibrate the signal light image in the image, and then the color histogram of the calibrated image is analyzed to calculate the color discrimination coefficient and use its rules to obtain the traffic light status. It can quickly and effectively judge the real-time status of traffic lights, which is helpful for intelligent vehicles to read the current traffic light information, and can be used to obtain the display status of traffic lights in intelligent driving, and has great application value in the field of intelligent driving .

Description

technical field [0001] The invention relates to an image processing method of a traffic signal lamp, in particular to a method for intelligently identifying the state of a circular traffic signal lamp. Background technique [0002] Traffic signal recognition, as an important part of assisted driving of unmanned vehicles, has received extensive attention. Circular traffic lights are the most common form of signal lights. Real-time and accurate identification of the status of circular traffic lights is conducive to the development of automotive assisted driving systems, and even the development of unmanned driving. Therefore, circular traffic light recognition has important research value and broad application prospects. Its practical significance is mainly reflected in the following three aspects: [0003] 1. Providing real-time intersection traffic information for unmanned vehicles is an indispensable part of the unmanned driving system. [0004] 2. It can be used as an a...

AI Technical Summary

Problems solved by technology

At present, a large number of recognition methods use template matching and support vector machines for classification operations, but both methods have shortcomings.

Template matching recognition is extremely sensitive to the established template, and a slight change in the recognition object may lead to unsuccessful template matching and recognition failure

The quality of the training samples will limit the performance of the support vector machine. In complex recognition scenarios, when the training is insufficient or the sample quality is not good, the support vector machine may lose its recognition ability

Images