Using Optical Flow in Motion Analysis for Evaluation of Active Music Therapy

In this study, we focused on the disabled children to introduce their voluntary motion. Motion analysis of their active music therapy is important for evaluation and planning, but the quantitative method is not well established yet. Since the therapists are too busy to spend time for analysis, more automatic tool is needed than the conventional motion analysis software. Therefore we investigated the use of optical flow in analyzing the client’s motion from the recorded video of music therapy. For this purpose, Lucas-Kanade method was used to calculate flow vectors. This algorithm needs feature points to be detected before flow calculation. Such features are limited to the ROI, that is, a part of the child’s body. Therefore mask image is introduced to specify the area for feature detection. OpenCV was used to develop experimental software and 12 video images of 5 subjects were processed. From the experimental results, the use of mask image to configure ROI was found to be effective. Though the correlation between results of optical flow and that of the conventional motion analysis was low in some cases, the timing of the detected motions was agreeable enough. These results may be effective for quantitative evaluation and intuitive visualization of active music therapy.