A gesture recognition method and system based on depth images

Abstract

The invention discloses a gesture recognition method based on a depth image. The depth images in a training data set and a test data set are collected by a depth sensor. The preset conditions of the nearest object of the sensor segment the gesture in the depth image; then obtain the projections of the gesture on three orthogonal planes, which are called the front projection, the side projection and the top projection; then extract the three The features of the contour fragments of each projection image are concatenated into the feature vector of the original depth gesture; finally, the classifier is trained to classify the gesture feature vector obtained from the depth image to be recognized, and the recognition result of the gesture to be recognized is obtained. The invention also provides a corresponding gesture recognition system. The gesture recognition method of the invention is simple and easy to implement, has strong popularization ability, high recognition accuracy rate, and can effectively overcome the influence of unfavorable factors such as messy background, illumination, noise and self-occlusion.

Description

Technical field [0001] The present invention belongs to the field of computer vision technology, and more specifically, relates to a gesture recognition method and system based on depth images. Background technique [0002] Gesture recognition has received attention due to its wide application in virtual reality, sign language recognition and human-computer interaction (HCI, computer games). Despite a lot of preliminary work, the application of traditional vision-based gesture recognition methods in real life is still far from satisfactory. Due to the nature of optical sensing, the optical sensor-based method is sensitive to light conditions and cluttered background. Therefore, it is generally unable to robustly detect and track hands, which greatly affects the performance of gesture recognition. In order to provide more robust gesture recognition, one of the effective ways is to use other sensors to capture gestures and movements, such as data gloves. Unlike optical sensors, s...

AI Technical Summary

Problems solved by technology

However, it is very difficult to efficiently obtain local support surfaces

In addition, the classification accuracy of the H3DF-based gesture recognition method on complex large gesture datasets needs to be further improved

Images