Night vehicle tail lamp extraction method based on descending luminance verification

Abstract

The invention provides a night vehicle tail lamp extraction method based on descending luminance verification. According to the method, firstly, an image is subjected to preprocessing such as ROI (region of interest) segmentation, tail lamp broad threshold filtration and the like, the preprocessed image is subjected to highlight white area extraction and subjected to constraint filtration such as a length-width ratio and the like, and a suspected vehicle lamp object area in the image is obtained; then, the halo range of each suspected vehicle lamp object is determined and subjected to hierarchical development, and a hierarchical chart accurately covering the tail lamp halo area is obtained; finally, descending luminance verification calculation is performed on the hierarchical chart of halo, a vehicle lamp object with a high passing rate in verification calculation is reserved, and the vehicle lamp object with high reliability is obtained. The method is applicable to an image-based night vehicle detection system and has good tail lamp morphological adaptability and color adaptability, and the detection rate of a night vehicle detection method based on a tail lamp is increased, and the robustness of the night vehicle detection method based on the tail lamp is improved. The night vehicle tail lamp extraction method can be used for a night vehicle detection algorithm for the automatic driving and intelligent traffic field.

Description

technical field [0001] The invention belongs to the field of intelligent vehicle (Intelligent Vehicle) based on computer monocular vision, in particular to a method for extracting tail lights of vehicles at night based on brightness decreasing verification. Background technique [0002] Vehicle detection at night can be applied to self-driving cars, early warning systems, etc. Night vehicle recognition based on computer vision takes the real-time video of the monocular camera as input to recognize the vehicle in front and judge the vehicle status. In vehicle-mounted vehicle detection at night, the tail light is the only obvious feature of the vehicle on the road at night, so the tail light detection is the key to the night vehicle detection method. [0003] Vehicle taillight extraction methods are mainly divided into three categories: methods based on color threshold filtering, methods based on brightness and taillight shape filtering, and methods based on machine learning....

AI Technical Summary

Problems solved by technology

Therefore, the detection effect of such methods is poor, and, due to the high computational complexity, it is usually difficult to guarantee real-time

Images