Nighttime road vehicle detection method based on luminance variance characteristics

Abstract

The invention discloses a nighttime road vehicle detection method based on luminance variance characteristics. For vehicle detection of nighttime road monitoring, a road surveillance camera is used as video source input, threshold segmentation of images is carried out using a histogram two-peak method, and nighttime vehicle images are preprocessed based on the atmospheric scattering principle; then, characteristic difference is statistically processed according to the data of luminance variance of a vehicle light and reflected light in an image, the reflected light and the vehicle light in the image are distinguished using a learning method based on a decision tree, and the reflected light in the image is eliminated; and finally, nighttime vehicle light detection is realized. The accuracy rate of nighttime vehicle detection and the time performance of detection are improved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to nighttime road vehicle detection in a monitoring scene. Based on the brightness variance feature of the image, the nighttime road reflected light screening, segmentation and elimination method using a decision tree learning method reduces the impact of road surface reflection on nighttime vehicle light detection. Interference, improving the accuracy and time performance of the night vehicle detection method using vehicle lights. Background technique [0002] In recent years, the daytime road vehicle detection technology based on surveillance video has been relatively mature; and for nighttime road monitoring vehicle detection, because the vehicle light feature is generally used to realize vehicle detection, it is easily affected by the reflected light interference of the road surface at night, resulting in The nighttime vehicle detection accuracy of headlights is...

AI Technical Summary

Problems solved by technology

[0002] In recent years, the daytime road vehicle detection technology based on surveillance video has been relatively mature; and for nighttime road monitoring vehicle detection, because the vehicle light feature is generally used to realize vehicle detection, it is easily affected by the reflected light interference of the road surface at night, resulting in Nighttime vehicle detection accuracy of headlights is low

[0003] The vehicle detection algorithm for road monitoring at night is mainly divided into: (1) The detection method based on machine learning. This method mainly extracts some description features of the vehicle and implements the training and classification of image features. This method requires a relatively comprehensive description of the vehicle. features, and establish a sample library for training, the calculation of this process is too large, there may be time performance problems, and an important requirement of vehicle detection is to achieve real-time detection; (2) Vehicle detection method based on taillights, this method mainly Vehicle detection is achieved through the color information or morphological features of the taillights, but the use of color information for detection is limited to a fixed color threshold, and the robustness to environmental changes is too low; One of the conditions in the information to filter the headlights cannot completely detect the vehicles on the road, and the adaptability is poor; (3) The vehicle detection method based on the headlights, because the headlights are the most prominent feature of the vehicle, Use the brightness information, shape information and other characteristics of the headlights to detect the headlights. Compared with the method of machine learning to classify the features, the headlight detection algorithm has lower computational complexity and can better achieve real-time performance. At the same time, relatively For the taillight detection method, it is less limited to environmental changes and more adaptable to the environment

However, there is also a problem with the vehicle detection algorithm based on headlights, that is, the reflected light formed by the lights on the road will interfere with the detection of the lights and affect the correct detection of the lights

In the existing solution, the interference is screened out based on the shape difference between the headlight and the interference. This method can reduce the interference on the original basis, but it still cannot rule out some road reflection interference similar to the shape of the headlight.

Images