HOG (Histogram of Oriented Gradient) and Mean Shift algorithm-based indoor pedestrian detection and tracking method

Abstract

The invention discloses an HOG (Histogram of Oriented Gradient) and Mean Shift algorithm-based indoor pedestrian detection and tracking method and belongs to the technical field of pedestrian detection and tracking in indoor complex environments. The method includes the following steps of: positioning of a Nao robot and a human body; HOG algorithm feature extraction; pedestrian detection HOG as well as SVM training and verification; Mean Shift algorithm target tracking; and Nao robot platform-based verification. According to the indoor pedestrian detection and tracking method of the invention, the HOG algorithm and the Mean Shift algorithm are adopted, and therefore, pedestrians can be detected in real time, and detected pedestrians can be tracked. Compared with methods in an early stage which perform pedestrian detection through Haar features, the indoor pedestrian detection and tracking method of the invention has a higher detection success rate and can reach a detection success rate of 90% in an INRIA pedestrian image database where backgrounds are complex. The indoor pedestrian detection and tracking method of the invention has the advantages of real-time performance, accurate detection and good practicability.

Description

technical field [0001] The invention belongs to the technical field of human body detection and tracking in an indoor complex environment, specifically uses a Nao robot as a platform, adopts a detection and tracking algorithm, and realizes the purpose of intelligent, fast and real-time detection and tracking of a human body. Background technique [0002] With the rapid development of computer vision technology, while people enjoy the convenience it brings in their daily life, they are also researching and exploring artificial intelligence to give robots the functions of human vision, making it more convenient, real-time and faster , so that the robot can handle the problem itself. In recent years, with the rapid development of digital image processing, pattern recognition, artificial intelligence, machine learning and other technologies, as well as the urgent needs of intelligent monitoring, intelligent vehicles and intelligent security, pedestrian detection and tracking tec...

AI Technical Summary

Problems solved by technology

Therefore, the field of vision of the Nao robot is limited, and if the human body leaves the field of vision of the robot, it will not be able to detect

Moreover, the human body is a non-rigid body, and its posture varies widely. It is more easily affected by factors such as scale, pedestrian appearance, lighting, complex scenes, etc., which inevitably brings great challenges to pedestrian detection and tracking technology.

Images