A tracking method for soft nonlinear moving objects using correlation method with confidence

Author(s):  
S. Tsuruoka ◽  
T. Kunisada ◽  
W. Ohyama ◽  
T. Yoshikawa ◽  
K. Sekioka
2014 ◽  
Vol 687-691 ◽  
pp. 564-571 ◽  
Author(s):  
Lin Bao Xu ◽  
Shu Ming Tang ◽  
Jin Feng Yang ◽  
Yan Min Dong

This paper proposes a robust tracking algorithm for an autonomous car-like robot, and this algorithm is based on the Tracking-Learning-Detection (TLD). In this paper, the TLD method is extended to track the autonomous car-like robot for the first time. In order to improve accuracy and robustness of the proposed algorithm, a method of symmetry detection of autonomous car-like robot rear is integrated into the TLD. Moreover, the Median-Flow tracker in TLD is improved with a pyramid-based optical flow tracking method to capture fast moving objects. Extensive experiments and comparisons show the robustness of the proposed method.


Atomic Energy ◽  
1996 ◽  
Vol 81 (1) ◽  
pp. 475-478
Author(s):  
A. F. Serov ◽  
S. V. Ursul ◽  
K. M. Kirillov ◽  
I. V. Padyukov ◽  
S. V. Krotov

2020 ◽  
Author(s):  
Jiankun Wu ◽  
Mingxuan Chen ◽  
Rui Qin ◽  
Feng Gao ◽  
Linye Song

<p>The objective extrapolation forecast is the main method for 0-1 hour convective storm nowcasting. Radar echo extrapolation was performed by using the 6 minute interval radar mosaics obtained from the radar images of 8 multi-radars in Beijing-Tianjin-Hebei region. A comparative study of two extrapolated forecasts of eighteen typical convective precipitation events occurred in Beijing-Tianjin-Hebei region from 2016 to 2018 was conducted. Compared with the tracking radar echoes by correlation method, the variational echo tracking method utilizes variational technique to compute the motion vector fields, and uses two strict constraints to get a better motion vector field. The results indicated that the variational echo tracking method performed better in prediction of the radar echo pattern, echo location, and echo intensity at 30- and 60-min forecast lead times: 1) A comparative study of the two extrapolated forecasts of four precipitation events in Beijing-Tianjin-Hebei region was conducted. The result indicated that the radar echo location, the echo pattern and echo intensity produced by the variational echo tracking method were closer to the real observation within one hour. 2) Quantitative evaluation for the two extrapolated forecasts of the eighteen typical convective precipitation events was conducted. Compared with the tracking radar echoes by correlation method, the probability of detection and the critical success index of the 30- or 60-min extrapolated forecast produced by the variational echo tracking method were higher, meanwhile the false alarm rate was lower when the radar echo threshold was 35dBz and 45dBz. Also, a quantitative evaluation classified by the weather type indicated that the variational echo tracking method performed better than the tracking radar echoes by correlation method in most weather types.</p>


2015 ◽  
Vol 734 ◽  
pp. 600-603
Author(s):  
Zhi Hai Sun ◽  
Bin Hu ◽  
Ying Meng ◽  
Wen Hui Zhou

Visual object detection and tracking have become an important step between computer vision and video analysis. Recent methods almost use mean shift for tracking problems, which are difficult to overcome the shortcoming with the initial object model. Initial object model is almost initialized manually by user, which is not smart enough and inconvenient. This paper considers the integration strategy for elliptical subtractive clustering and mean shift, and proposes a novel tracking method based on elliptical subtractive clustering.


2019 ◽  
Vol 50 (1) ◽  
pp. 27-32
Author(s):  
Yingwu Fang

The objective is to present an adaptive tracking approach of visual objects in the framework of particle filters (PF) and conflict redistribution strategy based on Dezert-Smarandache theory (DSmT). A combination rule of conflict redistribution was introduced into the researches on visual objects tracking, position and color information of moving objects were combined by modified combination strategy information, and the objects tracking model of fusing multiple information on the condition of occluded objects was established effectively. On the basis of building the execution panel of objects tracking simulation platform, many tracking experiments in complicated scene were carried out to validate the correctness and availability of the introduced method by comparing with other tracking method. Results showed that the introduced approach had excellent adaptive ability for dealing with high conflict among evidences under different occlusion condition, and could solve partly high-level vision tracking problems efficiently.


Sign in / Sign up

Export Citation Format

Share Document