지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학위논문
- 저자정보
- 발행연도
- 2013
- 저작권
- 한경대학교 논문은 저작권에 의해 보호받습니다.
이용수0
이 논문의 연구 히스토리 (2)
- 이 논문의 후속연구가 궁금하신가요?
- 연관 학술논문 또는 학술발표를 통해 보다 발전된 연구결과를 확인하실 수 있습니다.
- 이 논문의 연구 히스토리 확인하기
초록· 키워드
오류제보하기
A newly developed fall recognition system using the HMM
(Hidden Markov Model) with 3-axis accelerometer and video for the
elderly or patients is proposed. For this fall recognition system, a
new method for extracting fall feature parameters from 3-axis
acceleration and video is proposed. To distinguish between fall and
normal activities of daily living (ADL), the fall feature parameters
are applied to HMM.
To solve the problems such as deviation of interpersonal falling
behavioral patterns and similar fall actions, the proposed
HMMtraining method is used. The results which apply the
previously used parameters and newly developed parameters to the
HMM are compared and analyzed.
Compared to the conventional parameter SVM (Sum Vector
Magnitude) ASVM., the newly defined parameter GSVM (Gravity
weighted Sum Vector Magnitude) AGSVM of 3-axis accelerator gets
the best result to detect a fall. The results show that AGSVM can
distinguish various types of fall from ADLs with 100% sensitivity
and 97.96% specificity. Sensitivity and specificity of AGSVM are
improved to more 6% and 4% than those of ASVM, respectively.
In video field, to extract fall feature parameters from video, PCA
(Principal Component Analysis) of optical flow is used. As in case
of 3-axis acceleration data, the newly developed parameters applied
to the HMM are compared and analyzed. These fall feature
parameters used as HMM input data. The angle parameter V?(t)
among various fall feature parameters is best. The results show
that V?(t) can distinguish various types of fall from ADLs with
sensitivity of 91.5% and specificity of 88.01%.
(Hidden Markov Model) with 3-axis accelerometer and video for the
elderly or patients is proposed. For this fall recognition system, a
new method for extracting fall feature parameters from 3-axis
acceleration and video is proposed. To distinguish between fall and
normal activities of daily living (ADL), the fall feature parameters
are applied to HMM.
To solve the problems such as deviation of interpersonal falling
behavioral patterns and similar fall actions, the proposed
HMMtraining method is used. The results which apply the
previously used parameters and newly developed parameters to the
HMM are compared and analyzed.
Compared to the conventional parameter SVM (Sum Vector
Magnitude) ASVM., the newly defined parameter GSVM (Gravity
weighted Sum Vector Magnitude) AGSVM of 3-axis accelerator gets
the best result to detect a fall. The results show that AGSVM can
distinguish various types of fall from ADLs with 100% sensitivity
and 97.96% specificity. Sensitivity and specificity of AGSVM are
improved to more 6% and 4% than those of ASVM, respectively.
In video field, to extract fall feature parameters from video, PCA
(Principal Component Analysis) of optical flow is used. As in case
of 3-axis acceleration data, the newly developed parameters applied
to the HMM are compared and analyzed. These fall feature
parameters used as HMM input data. The angle parameter V?(t)
among various fall feature parameters is best. The results show
that V?(t) can distinguish various types of fall from ADLs with
sensitivity of 91.5% and specificity of 88.01%.
목차
등록된 정보가 없습니다.
최근 본 자료
댓글(0)
0