Object Tracking via Partial Least Squares Analysis
We propose an object tracking algorithm that learns a set of appearance models for adaptive discriminative object representation. In this paper, object tracking is posed as a binary classification problem in which the correlation of object appearance and class labels from foreground and background is modeled by partial least squares (PLS) analysis, for generating a low-dimensional discriminative feature subspace. As object appearance is temporally correlated and likely to repeat over time, we learn and adapt multiple appearance models with PLS analysis for robust tracking. The proposed algorithm exploits both the ground truth appearance information of the target labeled in the first frame and the image observations obtained online, thereby alleviating the tracking drift problem caused by model update. Experiments on numerous challenging sequences and comparisons to state-of-the-art methods demonstrate favorable performance of the proposed tracking algorithm.
Available from: Debarati Chakraborty
- "This problem has been studied over decades and there exist several literature (Yilmaz et al. 2006; Maggio and Cavallaro 2010). Some of the approaches are partially supervised (Comaniciu et al. 2003; Wang et al. 2012), that is, initial object/ background model is known and some of them are unsupervised (Heikkila and Pietikainen 2006; Pal and Chakraborty 2013). One may note that in video tracking the complete information is not available. "
[Show abstract] [Hide abstract]
ABSTRACT: This paper deals with several new methodologies and concepts in the area of rough set theoretic granular computing which are then applied in video tracking. A new concept of neighborhood granule formation over images is introduced here. These granules are of arbitrary shapes and sizes unlike other existing granulation techniques and hence more natural. The concept of rough-rule base is used for video tracking to deal with the uncertainties and incompleteness as well as to gain in computation time. A new neighborhood granular rough rule base is formulated which proves to be effective in reducing the indiscernibility of the rule-base. This new rule-base provides more accurate results in the task of tracking. Two indices to evaluate the performance of tracking are defined. These indices do not need ground truth information or any estimation technique like the other existing ones. All these features are demonstrated with suitable experimental results.
Available from: jsoftware.us
- "PLS analysis is a statistical method for modeling relations between sets of variables, the observed data is assumed to be generated by a process driven by a small number of latent variables. In , object tracking problem is formulated as a classification with PLS analysis to lean a low dimensional and discriminative feature subspace. Let m XR be an m-dimensional observed variations features space and n YR be an n-dimensional space of labels variables. "
Available from: PubMed Central
- "For example, in the field of object recognition, in 2011, Rosenfeld and Weinshall  proposed an algorithm to extract a foreground mask and to identify the locations of objects in the image. In the field of object tracking, in 2012, Wang et al.  used partial least squares (PLS) analysis to label the foreground and background of an image and the results showed that the proposed tracking algorithm was very powerful with the labeled foreground. In the field of content-based image retrieval, in 2006, Shekhar and Chaudhuri  investigated the influence of the foreground. "
[Show abstract] [Hide abstract]
ABSTRACT: A novel algorithm for automatic foreground extraction based on difference of Gaussian (DoG) is presented. In our algorithm, DoG is employed to find the candidate keypoints of an input image in different color layers. Then, a keypoints filter algorithm is proposed to get the keypoints by removing the pseudo-keypoints and rebuilding the important keypoints. Finally, Normalized cut (Ncut) is used to segment an image into several regions and locate the foreground with the number of keypoints in each region. Experiments on the given image data set demonstrate the effectiveness of our algorithm.
Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.