In order to recognize dynamic hand gestures with an effective and intelligent manner, this study proposes an integrated dynamic hand gesture recognition model based on the improved DTW (Dynamic Time Warping) algorithm that has a significant impact on the efficiency of dynamic trajectory analysis. The proposed model is divided into three operational steps: 1) using the three-frame difference method to track the gesture motion area by separating the dynamic motion region from the skin-like background; 2) applying the Hu-moment method to locate feature points from the extracted gesture motion region, and then use them to describe the motion trajectory; 3) exploring an improved DTW algorithm for gesture template match. Traditional DTW algorithm has excessive computational complexity and low operational efficiency. This study improved the running efficiency of DTW without reducing matching accuracy. Benchmarking experiments carried out on identifying the six classic dynamic gestures have yielded satisfactory classification results.
|Conference||25th IEEE International Conference on Automation and Computing|
|Abbreviated title||ICAC 2019|
|Period||5/09/19 → 7/09/19|