Mean Shift Trackers with Cross-Bin Metrics

Leichter, Ido

doi:10.1109/tpami.2011.167

Cited by 72 publications

(43 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Wang et al [47] incorporated shape knowledge into the appearance model of kernel based trackers, etc. Some other similarity measure criteria are utilized in kernel tracking; Yang et al [48] measured the similarity in average separation criterion in cluster analysis and Leichter [49] employ the Cross-bin metrics in mean-shift tracking. Minwoo et al [50] combined the mean-shift and Belief Propagation (BP) for multi-target tracking and the adaptive binning color model was utilized in Ref.…”

Section: Kernel-based Mean-shiftmentioning

confidence: 99%

Review on Kernel based Target Tracking for Autonomous Driving

Wang

Yue

Dong

et al. 2016

Journal of Information Processing

View full text Add to dashboard Cite

Significant progress has been made in the field of autonomous driving during the past decades. However, fully autonomous driving in urban traffic is still extremely difficult in the near future. Visual tracking of vehicles or pedestrians is an essential part of autonomous driving. Among these tracking methods, kernel-based object tracking is an effective means of tracking in video sequences. This paper reviews the kernel theory adopted in target tracking of autonomous driving and makes a qualitative and quantitative comparison among several well-known kernel based methods. The theoretical and experimental analysis allow us to conclude that the kernel based online subspace learning algorithm achieves a good trade-off between the stability and real-time processing for target tracking in the practical application environments of autonomous driving. This paper reports on the result of evaluating the performances of five algorithms by using seven video sequences.

show abstract

Section: Kernel-based Mean-shiftmentioning

confidence: 99%

Review on Kernel based Target Tracking for Autonomous Driving

Wang

Yue

Dong

et al. 2016

Journal of Information Processing

View full text Add to dashboard Cite

show abstract

“…It is employed to derive the object candidate that is the most similar to a given model by comparing the histogram of the object model in the current frame and the histogram of object candidate in the next frame, until finding maximum similarity between histograms, which are defined as the Bhattacharya coefficient, yet, algorithms based on mean shift may converge to a local maximum, and they are also sensitive to occlusions and objects with quick motion. A large number of improved algorithms [10,11] based on Mean Shift, such as CAMSHIFT (Continuously Adaptive Mean Shift) that can handle dynamically changing color distribution by adapting the search window size and computing color distribution to a search window, are proposed.…”

Section: Introductionmentioning

confidence: 99%

Visual Tracking Using Particle Filter Based on Gabor Features

Lahraichi¹,

Housni²,

Mbarki³

2018

IJIES

View full text Add to dashboard Cite

Abstract:Object tracking is a crucial step in video analysis, as well as detection and recognition. Particle filter, based on color histogram, is considered among multiple approaches that prove their effectiveness in this domain. It is apparent that if an object and its background, or more objects to be tracked, having nearly the same color histogram, the quality of tracking will be negatively affected. Therefore, combining color histogram and texture features for pixels in motion is considered a potential solution to this issue. In this paper, we propose a particle filter tracking method consisting of two stages: at the prediction stage, the state space model is represented by Box-Muller transform, and at the correction stage, the observation likelihood model is calculated in three steps: firstly, the background subtraction is used to extract the moving objects in order to reduce the impact of background pixels on color histogram. Secondly, the texture Gabor features are computed using moving pixels to distinguish between objects having the same color histogram. Finally, the histogram and texture features are combined to define a likelihood observation model. The proposed method is implemented using several video sequences. The visual and numerical results obtained are compared with those of traditional particle filter method, and show that our method is very efficient particularly in case of similar objects, and clearly improves the quality of tracking.

show abstract

“…This model relies on the universal rules for human language which are named as Zipfs law and word burstiness. Because it uses an algorithm instead of functional mapping to dynamically delineate the feature spaces, FSL is quite different from the mean-shift algorithm, which is another feature space model in image segmentation and video tracking (Comaniciu and Meer 2002;Leichter 2012). This model combines prior information and an assumption of consistency, which could not only embed the labeled information in similarity measurements, but also guide the clustering procedures.…”

Section: Introductionmentioning

confidence: 99%

Feature space learning model

Guan

Wang

Marchese

et al. 2018

J Ambient Intell Human Comput

View full text Add to dashboard Cite

With the massive volume and rapid increasing of data, feature space study is of great importance. To avoid the complex training processes in deep learning models which project original feature space into low-dimensional ones, we propose a novel feature space learning (FSL) model. The main contributions in our approach are: (1) FSL can not only select useful features but also adaptively update feature values and span new feature spaces; (2) four FSL algorithms are proposed with the feature space updating procedure; (3) FSL can provide a better data understanding and learn descriptive and compact feature spaces without the tough training for deep architectures. Experimental results on benchmark data sets demonstrate that FSL-based algorithms performed better than the classical unsupervised, semi-supervised learning and even incremental semi-supervised algorithms. In addition, we show a visualization of the learned feature space results. With the carefully designed learning strategy, FSL dynamically disentangles explanatory factors, depresses the noise accumulation and semantic shift, and constructs easy-to-understand feature spaces.

show abstract

Mean Shift Trackers with Cross-Bin Metrics

Cited by 72 publications

References 27 publications

Review on Kernel based Target Tracking for Autonomous Driving

Review on Kernel based Target Tracking for Autonomous Driving

Visual Tracking Using Particle Filter Based on Gabor Features

Feature space learning model

Contact Info

Product

Resources

About