Embedding holistic appearance information in part-based adaptive appearance model for robust visual tracking

• Published in: Electronics letters Vol. 49; no. 19; pp. 1219 - 1221
• Main Authors: Zeng, F.X, Huang, Z.T, Ji, Y.F
• Format: Journal Article
• Language: English
• Published: Stevenage The Institution of Engineering and Technology 12.09.2013; Institution of Engineering and Technology
• Subjects: Applied sciences; Artificial intelligence; Bayes methods; blooming discriminative trackers; Computer science; control theory; systems; Data processing. List processing. Character string processing; embedding holistic appearance information; Exact sciences and technology; Image and vision processing and display technology; image classification; image representation; Memory organisation. Data processing; naive Bayes classifier; object representation; object tracking; part‐based adaptive appearance model; Pattern recognition. Digital image processing. Computational geometry; robust visual tracking; Software; sparse multiscale Haar‐like features; image classification; naive Bayes classifier; object representation; blooming discriminative trackers; image representation; part-based adaptive appearance model; object tracking; embedding holistic appearance information; robust visual tracking; Bayes methods; sparse multiscale Haar-like features; Bayes estimation; Forgetting; Computer vision; Tracking; Motion estimation; Image processing; Updating; Modeling; Adaptive method; Experimental result; Model matching; Supervised learning; Multiscale method; Holism; Classification; Sparse representation; Probability learning
• ISSN: 0013-5194; 1350-911X; 1350-911X
• DOI: 10.1049/el.2013.2603

Part-based adaptive appearance model has been extensively used in increasingly popular discriminative trackers. The main problem of these methods is the stability plasticity dilemma. Embedding holistic appearance information in the part-based appearance model which is learned online to alleviate this problem is proposed. Specifically, the object is represented by sparse multi-scale Haar-like features and the appearance model is constructed with a naive Bayes classifier. Unlike the conventional methods, the classifier is trained by positive and negative samples that are weighted according to their similarity with the holistic appearance model, which is kept constant during the updating procedure. The constant holistic appearance information providing some constraints when updating the part-based appearance model makes the tracker more stable. The online updating procedure of the part-based appearance model makes the tracker adaptive enough to appearance changes. Experimental results demonstrate the superior performance of the proposed method compared with several state-of-art algorithms.