Region-of-interest filtering method and device for vehicle-mounted thermal imaging pedestrian detection

Abstract

The invention discloses a RoIs filtering method and device for vehicle-mounted thermal imaging pedestrian detection. The RoIs filtering method refers to the method of designing a three-layer cascade filter to filter non-pedestrian RoIs: the first layer filters out RoIs with abnormal sizes by calculating the pedestrian pixel height and the RoIs aspect ratio and setting a corresponding threshold interval; the second layer Calculate the vertical distance between its upper and lower boundaries and the current image road reference for each RoIs, calculate the threshold based on the pixel height of the RoIs, and filter out the RoIs with abnormal positions; the third layer searches possible according to the brightness vertical projection difference curve of each RoIs In the pedestrian head area, compare the difference between the Haar-like features of the head area and the adjacent background area, and filter out the RoIs missing the pedestrian head. This method can reduce the computational overhead of pedestrian detection and enhance the scene adaptability of the classifier on the premise of taking into account the accuracy of pedestrian detection. The RoIs filtering device includes an abnormal size RoIs filter, an abnormal position RoIs filter and a head-missing pedestrian RoIs filter.

Description

Technical field [0001] The present invention relates to pedestrian detection, and more specifically, to a method and device for filtering regions of interest (RoIs) for vehicle-mounted thermal imaging pedestrian detection. Background technique [0002] Vehicle thermal imaging pedestrian detection technology refers to the use of infrared thermal imaging cameras as visual sensors to capture images / videos of vehicle traffic scenes, and the use of machine learning methods on computers or embedded platforms to identify all pedestrian targets in the images / videos. The coordinate information of the smallest bounding rectangle identifies the position of each pedestrian on the image. [0003] This process consists of two key stages: RoIs extraction and RoIs classification and detection. The important factors affecting the computational cost and accuracy are the number of extracted RoIs and the performance of the classifier used. In the RoIs extraction process, in order to meet the requirem...

AI Technical Summary

Problems solved by technology

[0012] The purpose of the present invention is to provide a RoIs filtering method and device for vehicle-mounted thermal imaging pedestrian detection, aiming to promote the solution of problems such as decreased accuracy caused by computing performance bottlenecks and inability to meet real-time performance

Images