Method and device for eliminating shadows of vehicles

Abstract

The invention discloses a vehicle shadow eliminating method, comprising the following steps of: collecting a real-time video of a fixed area at a traffic intersection; if a movement prospect exits in a current frame image, obtaining a prospect mask block mass of a vehicle in the current frame image, and using a pre-set illumination model to separate a first shadow mask block mass of the vehicle from the prospect mask block mass; determining a shadow overlap ratio of the first shadow mask block mass and the illumination model; if the shadow overlap ratio is more than a first set threshold value, determining a shadow region of the vehicle in an RGB (Red, Green and Blue) color space; correcting the pre-set illumination model according to the shadow region and utilizing the corrected illumination model to separate a second shadow mask block mass of the vehicle from the prospect mask block mass; determining the shadow overlap ratio of the second shadow mask block mass and the corrected illumination model; and outputting a vehicle mask block mass after shadow separation until the shadow overlap ratio is less than or equal to the first set threshold value. The invention further discloses a vehicle shadow eliminating device.

Description

technical field [0001] The invention relates to the technical field of vehicle detection, in particular to a method and device for eliminating vehicle shadows. Background technique [0002] With the continuous development of science and technology, technical fields such as intelligent monitoring and intelligent transportation have developed vigorously. Video detection is widely used in intelligent transportation systems, especially vehicle detection at traffic intersections, vehicle tracking and vehicle driving behavior recognition. The basis of video detection is Vehicle detection, and the projected part of the shadow will change the shape and size of the detected vehicle target, affecting vehicle classification, tracking and behavior recognition. Eliminating shadows is a key issue in video detection. [0003] At present, shadow detection methods are mainly divided into two types: model-based and feature-based detection methods. The model-based method mainly uses the scene...

AI Technical Summary

Problems solved by technology

[0004] However, the above two shadow detection methods based on model and feature have disadvantages such as high algorithm complexity and slow calculation speed, which makes it difficult to achieve real-time processing of standard definition video.

Images