검색 상세

가속도 센서를 이용한 동적 손 제스쳐 인식

Dynamic Hand Gesture Recognition with Accelerometer

김상기 (Sangki Kim, 포항공과대학교 컴퓨터공학과)

원문보기

초록 moremore
Keypad based input method has been widely used for mobile device, such as mobile phone, MP3 player and remote control of electronic devices. As devices has more functions, keypad become more complex. Demand for intuitive interface and recent improvement of sensor and embbedded system technology ...
Keypad based input method has been widely used for mobile device, such as mobile phone, MP3 player and remote control of electronic devices. As devices has more functions, keypad become more complex. Demand for intuitive interface and recent improvement of sensor and embbedded system technology enable gesture-based interface for mobile device. There are some research interests in gesture recognition with mobile device that contains accelerometer and gyroscope. More atention has also been paid to gesture recognition with an accelerometer only, due to the relatively bulky size and high price of a gyroscope. Isolated gesture recognition is a kind of gesture recognition problem, which is to recognize single gesture from given data. It reqires a kind of pre-processing step to identify block of data involving single gesture before recognition process, whereas continuous gesture recognition system does not. This thesis has three main contribution on accelerometer based gesture recognition. First, we suggested an accelerometer based isolated gesture recognition system. Gravity acceleration components are removed form 3-d acceleration signals. Then,blocks of acceleration signals, involving single gesture, are extracted. Extracted blocks are applied to gesture model, to calculate likelihoods. Likelihoods for each gesture models are adjusted by corresponding weights. Weighted values of likelihood are finally used to determine a gesture model that best matches the input data in the recognition system. Hidden Markov models (HMMs) are used to model each gesture. In contrast other HMM-based approach, we use continuous LR HMMs and modified recognition criterion istead of discrete ergodic HMMs and maximu likelihood criterion. Second, we proposed an accelerometer based continuous gesture recognition system. Our HMMs-based isolated gesture recognition system can be exteneded continuous gesture recognition system by merging each HMMs to one huge HMM. This approach can be seen as 2-layered HMMs, the 1st-layer modeled the transition between gesture models, and 2nd-layer modeled each gesture. Maximum likelihood criterion, used in isolated gesture recognition system, can not be used to predict newly incoming acceleration signals in continuous gesture recognition system. Instead of the criterion, ??-based approach is used to predict test data. Finally, we suggest pseudo velocity signals, which is transformed from acceleratioin signals, for improvement of discriminative power. If we transform acceleration signals to other signals which represent velocity, including direction and speed of motion, it will be helpful to recognize the gesture. Accumulation of acceleration signals can be used for approximation of velocity signals. But in this case, noise of acceleration signals is also accumulated. As noise level of acceleration signal is high, due to lack of rotation information and imperfect gravity component removal, noise level of estimated velocity is extrimly high. We adapt template based approach, which is free from noise accumulation.
목차 moremore
1 Introduction 1
1.1 gesture-based interaction . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Accelerometer based gesture recognition . . . . . . . . . . . . . . . 2
...
1 Introduction 1
1.1 gesture-based interaction . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Accelerometer based gesture recognition . . . . . . . . . . . . . . . 2
1.3 Remote contoller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Related Work 7
2.1 Finite state machine based approach . . . . . . . . . . . . . . . . . 7
2.2 Template based approach . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Model based approach . . . . . . . . . . . . . . . . . . . . . . . . . 10
3 Isolated Gesture Recognition 14
3.1 System overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Gravity Acceleration Removal . . . . . . . . . . . . . . . . . . . . . 15
3.3 Gesture Block Extraction . . . . . . . . . . . . . . . . . . . . . . . . 18
3.4 Gesture Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.5 Weighting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.6 Early Decision . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.7 Experiments and Results . . . . . . . . . . . . . . . . . . . . . . . . 23
4 Additional Issues in Isolated Gesture Recognition 29
4.1 LR HMMs vs. Ergodic HMMs . . . . . . . . . . . . . . . . . . . . . 29
4.2 Recognition criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3 experimental result . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5 Continuous Gesture Recognition system 36
5.1 2-layered HMMs for continuous gesture recognition . . . . . . . . 37
5.2 Hierachical Hidden Markov Model . . . . . . . . . . . . . . . . . . 39
5.3 ??-based gesture recognition . . . . . . . . . . . . . . . . . . . . . . 40
5.4 Experimental Result . . . . . . . . . . . . . . . . . . . . . . . . . . 41
6 Pseudo Velocity for accelerometer-baseed gesture recognition 44
6.1 Gestures with remote control . . . . . . . . . . . . . . . . . . . . . 45
6.2 Pseudo velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
6.2.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
6.2.2 Procedure to generate pseudo velocity signal . . . . . . . . 46
6.2.3 Normalization . . . . . . . . . . . . . . . . . . . . . . . . . . 47
6.2.4 Comparison function . . . . . . . . . . . . . . . . . . . . . . 50
6.2.5 Direction templates . . . . . . . . . . . . . . . . . . . . . . . 54
6.3 Zero Velocity Compensation . . . . . . . . . . . . . . . . . . . . . . 56
6.4 HHMs based classifier . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.5 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7 Conclusion 66
Bibliography 70
한글 요약문 75