A Color Image Oriented Vehicle Detection Method

Abstract

A color image-oriented vehicle detection method is disclosed, which has strong ability to describe the appearance of objects to be detected, strong adaptability, moderate training samples required, and low time consumption. The method includes the following steps: (1) constructing a dictionary set with category discrimination ability for vehicle category and non-vehicle category through feature design of vehicle detection and hierarchical sparse-dense dictionary set modeling; (2) sliding window scanning and window Feature extraction; window classification based on sparse-dense dictionary set; duplicate detection box removal based on kernel density estimation non-maximum suppression to filter highly overlapping detection results.

Description

technical field [0001] The invention relates to the technical field of computer vision and pattern recognition, in particular to a color image-oriented vehicle detection method. Background technique [0002] As the basic technology of computer vision, vehicle detection technology has important application value in unmanned driving system assisted by color camera and urban traffic monitoring system. Existing target detection methods can be roughly divided into the following three categories according to the models used: [0003] The first type of detection method is classification detection based on sliding window. The method generally uses a sliding window scanning process to classify each window based on a general machine learning classification model without deep learning to achieve detection. Viola and Jones proposed a face detection method based on local pixel difference features and AdaBoost classifier; Dalal and Triggs proposed a pedestrian detection method combining...

AI Technical Summary

Problems solved by technology

This type of method will have a high time consumption due to the processing of the 3D model

Images