A Two-Handed Gesture Recognition Technique on Mobile Devices Based on Improved DTW Algorithm

The majority of traditional gesture recognition relies on cameras, easily affected by environmental noises. Moreover, most of them are one-handed gestures, whose identifying speed and accuracy are limited. Therefore, this paper proposed a two-handed gesture recognition technology based on improved dynamic time warping (DTW) algorithm and common mobile devices. The data are collected by common carry on mobile communication devices instead of wearable devices. By constructing boundary linked list, traditional DTW algorithm is optimized, so we realized two-handed gesture trajectory recognition. The results show that, under the prerequisite of guaranteeing accuracy, the method can considerably reduce the algorithm's computation complexity, and effectively improve the speed of recognition.