We present here a real-time tracking algorithm for thermal video from a sports game. Robust detection of people includes routines for handling occlusions and noise before tracking each detected person with a Kalman filter. This online tracking algorithm is compared with a state-of-the-art offline multi-target tracking algorithm. Experiments are performed on a manually annotated 2-minutes video ...

In this paper we discuss about a target tracking algorithm based on Mean Shift and Kalman Filter, which is suitable for high speed moving target tracking. The basic Mean Shift algorithm is described in this paper as well. Although basic Mean Shift algorithm can realize target tracking without arguments or searching all the areas effectively, it has shortcomings which can limit its effectiveness...

In this paper, we present an overview of several visual tracking methods for industrial augmented reality applications. We show that no universal algorithm can deal with the large number of possible scenes, and that the different methods have to be seen as complementary approaches that all have their strengths and weaknesses. The main difficulty, then, consists in combining existing building bl...

This paper presents a real-time visual tracking system in the robot soccer domain. The details of the Essex Wizards robot football team and its control system are described. A real-time visual tracking algorithm has been developed. Some initial results of the image pre-processing and visual tracking algorithms in real environment setting are given to show the performance of the system.

In most multi-object tracking algorithms, tuning of model parameters is of critical importance for reliable performance. In particular, we are interested in designing a robust tracking algorithm that is able to handle unknown false measurement rate. The proposed algorithm is based on coupling of two random finite set filters that share tracking parameters. Performance evaluation with visual sur...

− A Background Information Fusion (BIF) algorithm and its application in video target tracking is proposed in the paper. The BIF algorithm is based on the fact that there are redundant information between different frames. Unlike most of available tracking algorithms based on target features, which focus on the problems of what features the target have, and how does these features change when t...

The traditional approaches for formant frequency estimation are misled by spectral peaks in unvoiced speech and perform very poorly in transient background noise. Also, these traditional algorithms are not robust in real life noise environment and are unable to recover quickly after periods of silence. Robust formant tracking algorithm is discussed. Testing of the algorithm using various speech...

Image target tracking is an important part in computer vision field, and has a broad application prospect in modern society. Lack of online learning function, the traditional target tracking algorithm was based on the single track as long as possible. The target deformation and occlusion will lead to target tracking lost. TLD (tracking-learningdetection) is the long-term effective algorithm in ...