A Two-Step Motion Retrieval Method

Project: Research

View graph of relations

Description

In human motion animation, motion capture techniques are often used, as they provide an efficient way to produce realistic motion. However, as the motion database gets larger, we are faced with two problems. First, the motion database takes up a lot of memory, which can be critical for mobile devices. Second, with such a large database, the motion retrieval process becomes time consuming. In this project, we propose to address the first problem by compressing each motion file in the database. We propose to address the second problem by developing an efficient two-step method for retrieving motions from the compressed motion database. The first step is to quickly filter irrelevant motions based on eigen-comparison. The second step is to perform a detailed comparison on the selected candidate motions. We will conduct experiments to evaluate the performance of the proposed method.

Detail(s)

Project number7002768
Grant typeSRG
StatusFinished
Effective start/end date1/05/1214/10/15