Exploring Dance Movement Data Using Sequence Alignment Methods

Abstract

Despite the abundance of research on knowledge discovery from moving object databases, only a limited number of studies have examined the interaction between moving point objects in space over time. This paper describes a novel approach for measuring similarity in the interaction between moving objects. The proposed approach consists of three steps. First, we transform movement data into sequences of successive qualitative relations based on the Qualitative Trajectory Calculus (QTC). Second, sequence alignment methods are applied to measure the similarity between movement sequences. Finally, movement sequences are grouped based on similarity by means of an agglomerative hierarchical clustering method. The applicability of this approach is tested using movement data from samba and tango dancers.

Movement Pattern Analysis Based on Sequence Signatures

“…At the end of this section, we compare the proposed methodology with the one presented in [63] where the same case study (i.e., dance) was studied. We describe the strengths and weaknesses of the methodology compared to the one in [63] from data modeling point of view.…”

“…We describe the strengths and weaknesses of the methodology compared to the one in [63] from data modeling point of view. In our previous work [63], we identified similarities in the movement behavior between pairs of interacting moving objects over time by cross-pollinating QTC with sequence alignment methods (SAMs). There, sequence alignment was used to calculate the similarity between movement sequences of QTC relations for some reasons.…”

Pair-Activity Analysis From Video Using Qualitative Trajectory Calculus

“…More recently, a method for analyzing dance movement using QT C has been presented in [18]. The problem domain of this work corresponds more closely to ours and the method comprises three main components: transformation of movements into QT C sequences; sequence alignment; and comparison of sequences using the conceptual distance.…”

“…Our work directly addresses these limitations. Chavoshi et al [18] used a scoring system based on the conceptual distance, using both QT C B and QT C C representations. This is defined as the number of changes required to incrementally change one value to another; for example the distance between "-" and "+" is two, because the transition must pass through the intermediate state [40].…”

“…In their work on dance sequence analysis, Chavoshi et al used both DTW with real-valued features [25], and QT C with sequence alignment [18], as analysis tools. They present a discursive comparison of DTW and sequence alignment methods in [18], but favour sequence alignment for their QT C approach because it is easy to interpret visually, and can be used to align multiple sequences. We have also followed a sequence alignment approach in our own work, which we use with normalised QT C sequences.…”