A method for judging the opening state of vehicle high beams at night based on dual cameras

Abstract

The invention discloses a method for judging the open state of vehicle high beams at night based on dual cameras. The method includes the following steps: Step 1: Arrange two cameras in the device at the same time, the two cameras are in the same position, and the images captured by the two cameras are The pictures are synthesized into one picture, one is a low-exposure picture, and the other is a high-exposure picture; Step 2: Use the low-exposure picture to determine the turn-on state of the lights, and at the same time, track the vehicle's driving trajectory. The specific processing algorithm is as follows: 2.1: For the image Perform binarization processing, and then perform dilation processing on the binary image, and perform a labeling algorithm on the expanded binary image to find each connected domain. Through the implementation of the present invention, it is possible to correctly identify the open state of the high beam of the vehicle driving at night, and to correctly identify the license plate number of the vehicle, thereby improving the identification accuracy rate of the high beam to a large extent, and achieving Outstanding progress.

Description

technical field [0001] The invention relates to the technical field of high beam light discrimination, in particular to a method for discriminating the open state of vehicle high beam light at night based on dual cameras. Background technique [0002] The method of monitoring video is used to complete the identification of the high beam headlights of vehicles driving at night. At present, the most effective method is to set a front-facing video image of the driving vehicle from far to near, which brings a more difficult situation. , as the illuminant of the car lights, and the night environment constitute a wide dynamic range, so that a single camera cannot simultaneously complete the high beam status and the vehicle identity information (license plate) with the high beam turned on. , a single camera cannot obtain sufficient high beam feature information even with a fixed exposure parameter. [0003] Since the wide dynamic range formed after the lights are turned on cannot ...

AI Technical Summary

Problems solved by technology

[0002] Using surveillance video to complete the recognition of the high beams of vehicles driving at night, the most effective method at present is to set up video images that can capture the front of the vehicle from far to near, which brings a more difficult situation , the car lights as illuminants, and the night environment constitute a very wide dynamic range, making it impossible for a single camera to simultaneously complete the clear capture of the high beam status and the vehicle identity information (license plate) with the high beam turned on. On the other hand , a single camera cannot obtain sufficient high-beam feature information when the exposure parameters are fixed

[0003] Due to the wide dynamic range formed after the headlights are turned on, the physical characteristics of the current camera cannot adapt to it. As a result, if the exposure of the picture is adjusted to allow the light-emitting state of the high-beam and low-beam lights to be clearly seen, it is impossible to obtain a clear picture at all. License plate information, even if it is judged that there is an act of turning on the high beam in violation of regulations, the identity information (license plate number) of the vehicle cannot be determined; indistinguishable

[0004] Since the use of high beams in violation of regulations needs to be judged whether it continues to be on for a period of time, the multi-level exposure switching of a single camera will lead to discontinuity of information, making law enforcement evidence insufficient