Search

In this paper, we propose a sparse coding tracking algorithm based on the Cartesian product of two sub-codebooks. The original sparse coding problem is decomposed into two sub sparse coding problems. And the dimension of sparse representation is intensively enlarged at a lower computational cost. Furthermore, in order to reduce the number of L1-norm minimization, ridge regression is employed to exclude the substantive outlying particles according to the reconstruction error. Finally the high-dimension sparse representation is put into the classifier and the candidate with the maximal response is considered as the target. Both qualitative and quantitative evaluations on challenging benchmark image sequences demonstrate that the proposed tracking algorithm performs favorably against several state-of-the-art algorithms. The PSC under multi-task is more robust and stable than original PSC.SVM has better generalization under PSC than under SC.The PSCT has stronger discriminative ability than SCT under the same conditions.

The key problem of the adaptive mixture background model is that the parameters can adaptively change according to the input data. To address the problem, a new method is proposed. Firstly, the recursive equations are inferred based on the maximum likelihood rule. Secondly, the forgetting factor and learning rate factor are redeffned, and their still more general formulations are obtained by analyzing their practical functions. Lastly, the convergence of the proposed algorithm is proved to enable the estimation converge to a local maximum of the data likelihood function according to the stochastic approximation theory. The experiments show that the proposed learning algorithm excels the formers both in converging rate and accuracy.

in this paper, we address the problem of tracking drift and failure due to background clustering, illumination and scale changes. To resolve these problems, we propose an efficient model that project original RGB color space to a more robust color space—Color Names feature space. Furthermore, we represent objects by background weighted histograms, and thus suppress the similar background around. Moreover, we adapt a two-step tuning way to estimate the scale by coarse tuning with gradient ascent and fine tuning with constrained items. Back-forward scale check is used to ensure the precision of scale estimation. The results show that our approach is robust to background clustering, illumination variation, partial occlusion, and viewpoint changes.

A dynamic sensor management algorithm based on improved efficacy function is proposed to solve the multi-sensor faced to multi-target management problem. The tracking task precision requirements (PR), target priority (TP) and sensor used cost (SC) are considered to establish the efficacy function by weighted sums the normalized value of the three factors. The PR is described by the diversities of the desired covariance matrix and the predicted Covariance matrix which is obtained by the centralized sequential Kalman filtering algorithm. The TP model is established based on multi-attribute linear weighting and SC is estimated by practical experience. The simulation results proved that the proposed method can meet the requirements of PR and adjust sensor selection according to TP and SC. This makes sensor management scheme more reasonable and effective.