Gesture Recognition And Its Application Based On Depth Data

Human-computer interaction technology is a technology that people exchange information with computers through transmission devices.With the development of science and technology,the traditional human-computer interaction devices,such as mouse,handle,keyboard and so on,have gradually unable to meet the growing needs of people.Gesture recognition is a type of human-computer interaction that uses gestures to interact with machine.It has the characteristics of natural,convenience and easy to learn,gradually become the research focus of human-computer interaction.Both the gesture recognition technology based on the data glove and the gesture re cognition based on the machine vision have shortcomings: The data glove is expensive and inconvenient to use;the gesture recognition based on machine vision can solve the problems caused by the data glove,but it is easily disturbed by the illumination an d the background.Based on the above reasons,this dissertation chooses gesture recognition based on depth data to study.Depth data is captured by the Xtion depth device,which reflects the distance between each object in the target area and the camera.Depth data is not affected by illumination.It greatly reduces the interference of various environmental factors on gesture recognition,which can largely reduce the complexity of gesture recognition algorithm and improve its accuracy.The research content of this dissertation is to establish a gesture recognition system by using the depth data and to utilize the deep data information in every part of the system,finally realizing the application.Gesture recognition system is divided into three parts,in the part of gesture segmentation,gesture segmentation algorithm based on skin color information and depth data is improved,which improves the effectiveness and accuracy of the gesture segmentation algorithm.In the part of gesture tracking,a gesture tracking algorithm based on depth data is proposed.The tracking drift problem is improved on the basis of ensuring the effectiveness of the tracking algorithm.In the part of gesture recognition algorithm,the support vector machine(SVM)gesture recognition algorithm based on the Histogram of Oriented Gradient(HOG)is selected to classify and recognize 5 kinds of defined static gestures,and obtain good recognition accuracy.At the same time,we also use the two-dimensional location information and depth data of the target gesture to extend the gesture recognition from the traditional two-dimensional space to three-dimensional space.This method expands the application field of gesture recognition,which has great significance to the further development of gesture recognition.After all the above steps are completed,an application is established by using the established gesture recognition system so as to control the movement and operation of the mouse pointer and the switching of the page by the movement and change of the five gestures.The application is able to verify the validity and practicability of the gesture recognition system in this dissertation.The application also can be more specific to recognize the deficiency of the gesture recognition system.This can make the research more perfect,and the popularity of gesture recognition applications can be achieved soon.