Search

This paper presents a new behavior classification system to analyze human movements around a view circle using time-order similarity distributions. To maintain the view in-variance, an action is represented not only from its spatial domain but also its temporal domain. After that, a novel alignment scheme is proposed for aligning each action to a fixed view. With the best view, the task of behavior analysis becomes a string matching problem. One novel idea proposed in this paper is to code a posture using not only its best matched key posture but also other unmatched key postures to form various similarity distributions. Then, recognition of two actions becomes a problem of matching two time-order distributions which can be very effectively solved by comparing their KL distance via a dynamic programming scheme. [ABSTRACT FROM AUTHOR]

This paper proposes a novel dynamic sparsity-based classification scheme to analyze various interaction actions between persons. To address the occlusion problem, this paper represents an action in an over-complete dictionary to makes errors (caused by lighting changes or occlusions) sparsely appear in the training library if the error cases are well collected. Because of this sparsity, it is robust to occlusions and lighting changes. In addition, a novel Hamming distance classification (HDC) scheme is proposed to classify action events to various types. Because the nature of Hamming code is highly tolerant to noise, the HDC scheme is also robust to environmental changes. The difficulty of complicated action modeling can be easily tackled by adding more examples to the over-complete dictionary. More importantly, the HDC scheme is very efficient and suitable for real-time applications because no minimization process is involved to calculate the reconstruction error. [ABSTRACT FROM AUTHOR]

This paper proposes a novel system to analyze human-object interaction events happening between hands and faces in real time. Two challenging problems in this event analysis must be addressed, i.e. , there is no prior knowledge (like shape, color, size, and texture) about the handheld objects, and there are large spatial–temporal variations in event representation. For the first challenge, a novel ratio histogram is proposed to find important color bins to locate handheld objects and their trajectories via a code book technique. This scheme is different from other boosted methods which require very time-consuming estimations to search reliable body configurations. For the second challenge, a mixture of HMMs is proposed to describe an event not only from its dynamic context but also its multiplicity context. It can be performed in real time because an exhaustive search process is avoided to find possible interaction pairs between objects and body parts. [ABSTRACT FROM AUTHOR]