Research On Ring-type Virtual Keyboard Based On MEMS Sensor

Nowadays,human-computer interaction has become an indispensable part of People’s daily life.With the innovation and development of electronic devices,people are eager to find a more convenient,user-friendly and popular way of human-computer interaction to meet the needs of new features.In recent years,the emergence of virtual reality technology has stimulated many researchers to study and explore virtual keyboard technology,and showed an explosive growth.With existing technology,virtual keyboards can be roughly divided into three categories:wearable virtual keyboards,virtual keyboards based on machine vision and some emerging non-contact virtual keyboards.However,most wearable virtual keyboard hardware systems are too complex to use;In machine vision based virtual keyboard,the low efficiency of recognition algorithm has become a major constraint which make it difficult to be popularized on a large scale.Other contactless virtual keyboards are facing the shortcomings,such as low accuracy,poor stability and other potential problems.In this paper,we propose a ring-type virtual keyboard mainly based on MEMS motion sensor and attitude features,which has simple hardware structure and good performance in portability,ease of use,accuracy and algorithm efficiency.The main research content and results of this paper can be summarized as follows:Firstly,a ring virtual keyboard based on MEMS inertial measurement unit(IMU)is proposed and designed.The ring-type virtual keyboard uses wearable sensors to enter characters without physical contact with the keyboard.In this scheme,we adopt the ring embedded with MEMS motion sensor to collect and process the finger motion data.Through the analysis and data mining of raw motion data,the attitude estimation is carried out by angle complementary filtering algorithm,so as to extract features and classification,so as to realize the function of virtual keyboard.Secondly,we build the hardware platform according to the established scheme and then design and complete the test and verification of the corresponding algorithm function.It includes the overall design of ring-type ring,data collection and transmission.Also,we excavate the deep features of the motion data and synthesize various analysis methods,and finally carry out the feasibility study of the test scheme.The whole process includes data preprocessing,data segmentation,attitude estimation,feature extraction and feature selection and other important stages.The corresponding model file was established to determine the special mapping relationship between keystroke features and keys.Finally,different machine learning algorithms are used to realize the keystroke recognition in the realization of virtual keyboard.Finally,a new SCM algorithm based on multilevel decision is proposed and designed.And then we implemented the function of the algorithm on PC and MCU respectively for detail illustration of the characteristics in recognition accuracy,running speed and memory requirements.By comparing with other machine learning algorithms in the two different operating environments,we finally evaluate the characteristics of the algorithm comprehensively.Experimental result shows that the ring-type virtual keyboard scheme proposed in this paper can effectively accomplish keystroke recognition.With the combination of attitude angle and acceleration features,a high recognition accuracy can be achieved through the self-defined SCM algorithm and a fast recognition and classification can be implemented with less memory requirements and loose hardware conditions.