Extraction of Motion Characteristics Corresponding to Sensitivity Information Using Dance Movement

Abstract

This study is intended to quantitatively clarify the relationship between the motion characteristics behind the human motion in complicated motions like dancing and the subjective impressions of the observer. It examines the impression structures related to the motion of a determined body part of dancing and considers the motion characteristics giving a specific impression. To compare and consider the impression structures related to the motion of a body part, the authors made a principal component analysis, one of the multi-variable analytic methods, to check the arm and leg motions for any differences in the impression structure. Similarly, they considered any differences in the impression structures due to the experience knowledge of dance. Next, to consider any differences in the physical features that have effect on the impressions, they quantified the motion characteristics and used a heavy regression analysis to estimate the common motion characteristics that give the same impressions. In addition, they used the characteristics of the legs that are parts of the motion presumed to have the relationship with the impressions to reproduce the motion with CG for the consideration of these impressions. As a result, when the impressions of the arm and leg motions were compared, four impression evaluation axes of "like-dislike," "dynamic-static," "individual-monotonous," and "collected-wide" were extracted as the axes that evaluated the same impressions, but the impressions of "hard-soft" and "heavy-light" were extracted only from those of each arm or leg motion. When the evaluation axes of the impressions were compared between groups with differences in the knowledge of dance, five similar evaluation axes were extracted for each of them and there was no big difference in the impression structures themselves, but significant differences were found for the evaluation of impressions between the words used for the sensitivity evaluation in difference in knowledge. Attention was paid to the characteristics of the motion generating each impression to show the relationship between motion characteristics and subjective impressions.

Intelligent affect regression for bodily expressions using hybrid particle swarm optimization and adaptive ensembles

“…Most recent automatic recognition research makes use of either body posture or movement as the source of affective information whereas many early developments have focused on recognizing emotions from expressive dance (e.g. Camurri et al, 2003;Camurri et al, 2004;Park et al, 2004;Kamisato et al, 2004). Camurri and colleagues (Camurri et al, 2003;Camurri et al, 2004) extracted dynamic motion cues from dancers' body movements to differentiate between basic emotions.…”

Recognizing Affective Dimensions from Body Posture

“…In the affective computing field, models for the automatic recognition of discrete emotion categories from static body postures have been proposed [2][5] [15]. Other studies have shown that body motions extracted from dance sequences [3] [24] [14] can convey emotion in terms of discrete categories effectively. To our knowledge, what are missing still are models for the automatic recognition of affective dimensions from posture as there are for motion [4].…”

Expression of emotional states during locomotion based on canonical parameters

“…Studies following this idea differ methodological by either exposing viewers to real actors playing [23], video scene of actors playing [2], computer animations of virtual humans [26], [27], point-light animations that conceptualize human body movements [28], [29], [30], [31], [22], or drawn figures [32]. The results of these and similar studies show that affective states can be identified by observing static postures [33], [26], [34], [35] or moving behavior [36], [37], [23], [2].…”