System design for moving target tracking based on mean-shift algorithm

2018 ◽  
Author(s):  
Yu Yang ◽  
Chengpo Mu ◽  
Kai Liu
Keyword(s):  
System Design ◽  
Target Tracking ◽  
Mean Shift ◽  
Moving Target ◽  
Moving Target Tracking ◽  
Mean Shift Algorithm

Research on the Mean Shift Target Tracking and Recognition Technology Based on Multi-Feature Fusion

2013 ◽  
Vol 705 ◽  
pp. 561-564 ◽  
Author(s):  
Yong Jun Peng ◽  
Fen Tan ◽  
Jun Sun
Keyword(s):  
Target Tracking ◽  
Video Sequence ◽  
Feature Fusion ◽  
Mean Shift ◽  
Moving Target ◽  
Compressed Domain ◽  
Target Movement ◽  
Moving Target Tracking ◽  
Template Update ◽  
The Mean

This article topics based on multi-feature fusion the Mean shift target tracking technology belongs to the field of intelligent video analysis, moving target tracking is interested in moving target location each image in a video sequence to find and acquire the target movement. Moving target tracking problem can be stated as interested in moving target movement prediction in the video sequence, feature extraction, feature matching and template update problem. In this paper, we consider using compressed domain features as a complement of the color features to extract the compressed domain features first need to understand the compressed domain detection technology. Detection based on the compressed domain, that is, in the case of not decoding or a small amount of decoding, directly on the compression characteristics of the image analysis, in order to achieve the detection of the image moving object.

Research on Mean Shift Algorithm

2013 ◽  
Vol 756-759 ◽  
pp. 4021-4025 ◽  
Author(s):  
Yi Zhi Zhao ◽  
Huan Wang ◽  
Guo Cai Yin
Keyword(s):  
Computer Vision ◽  
Object Tracking ◽  
Target Tracking ◽  
Mean Shift ◽  
Moving Target ◽  
Object Model ◽  
Probability Density Estimation ◽  
Complete Occlusion ◽  
Mean Shift Algorithm ◽  
High Level

Computer vision is a diverse and relatively new field of study. Object tracking plays a crucial role as a preliminary step for high-level image processing in the field of computer vision. However, mean shift algorithm in the target tracking has some defects, such as: the application of fixed bandwidth for probability density estimation usually causes lack of smooth or too smooth; moving target often appears partial occlusion or complete occlusion due to the complexity of the background; background pixels in object model will induce localization error of object tracking, and so on. Therefore, this paper elaborates several elegant algorithms to solve some of the problems. After discussing the application of Mean shift in the field of target tracking, this paper presented an improved Mean shift algorithm by combining Mean Shift and Kalman Filter.

Sign in / Sign up

Export Citation Format

Share Document