Human motion retrieval based on freehand sketch

Abstract

In this paper, we present an integrated framework of human motion retrieval based on freehand sketch. With some simple rules, the user can acquire a desired motion by sketching several key postures. To retrieve efficiently and accurately by sketch, the 3D postures are projected onto several 2D planes. The limb direction feature is proposed to represent the input sketch and the projected-postures. Furthermore, a novel index structure based on k-d tree is constructed to index the motions in the database, which speeds up the retrieval process. With our posture-by-posture retrieval algorithm, a continuous motion can be got directly or generated by using a pre-computed graph structure. What's more, our system provides an intuitive user interface. The experimental results demonstrate the effectiveness of our method.

Sketch-based human motion retrieval via selected 2D geometric posture descriptor

“…What is more, some index structure such as K-D-tree or R-tree can applied to index the feature space, which can speed up the k-NN retrieval process. In this work, k-d tree is used as Tang [28] introduced. …”

“…As far as geometric feature concerned, some information like limb length and joint coordinate are removed directly because each user has different sketch style and the body proportion is hard to be unified for the same posture. The direction of the limbs, an intuitional features, is usually used [21,28] on sketch. However, only the direction of the limbs cannot character a motion sufficiently.…”

“…Inspired by Tang [28], we built a retrieval system which allows the user to define a motion by sketch several key postures. The posture sketch has some simple limits such as (a) one sketch posture includes five strokes; (b) the first stroke is body; (c) the view tilt and elevation angles are set to zero and a azimuth angle that makes the motion easily distinguished is selected.…”

“…Here the key posture extraction algorithm proposed by Tang [28] is used. Then the key postures are projected into eight 2D plains, orthographic projection is applied and eight projection directions are denoted as f D k !…”

Assessing similarity models for human‐motion retrieval applications

“…These applications require to process motion data effectively and efficiently. To do that, they usually exploit specialized techniques for annotating segments of motions by textual descriptions or retrieving motions or their sub‐motions that are similar to a query motion example . Such retrieval and annotation (classification) techniques require to process characteristic aspects of motion data, rather than raw spatio‐temporal 3D coordinates.…”