... Asian Jornal of Control , 12: 704713. [CrossRef] View all references), state estimation (Bal... more ... Asian Jornal of Control , 12: 704713. [CrossRef] View all references), state estimation (Balasubramaniam, Lakshmanan, and Jeeva Sathya Theesar 20101. Balasubramaniam, P, Lakshmanan, S and Jeeva Sathya Theesar, S. 2010. ... Yu, JJ, Zhang, KJ and Fei, SM. 2009. ...
... Cao, Jiangtao, Li, P. and Liu, Honghai (2009) Improved interval type-2 fuzzy logic controller... more ... Cao, Jiangtao, Li, P. and Liu, Honghai (2009) Improved interval type-2 fuzzy logic controller (Chinese). Control and Decision, 24 (10). ... Departments/Research Groups: Faculty of Creative and Cultural Industries > School of Creative Technologies. Depositing User: Trish Skinner. ...
An exemplar-based view-invariant human action recognition framework is proposed to recognize the ... more An exemplar-based view-invariant human action recognition framework is proposed to recognize the human actions from any arbitrary viewpoint image sequence. In this framework, human action is modelled as a sequence of body key poses (i.e., exemplars) which are represented by a collection of silhouette images. The human actions are recognized by matching the observed image sequence to predefined exemplars, in which the temporal constraints are imposed in the exemplar-based Hidden Markov Model (HMM). Furthermore, a new two-level recognition framework is introduced to improve the discrimination capability for the similar human actions. The aim of the first level recognition is to decide an equivalent set in which the testing action is included instead of directly achieving the final recognition results. In the second level, the weighted contour shape feature is used to calculate the observation probability to discriminate the similar actions. The proposed framework is evaluated in a public dataset and the results show that it not only reduces computational complexity, but it is also able to accurately recognize human actions using single cameras. Besides it is verified that the weighted contour shape feature is effective to differentiate the similar arm-related actions.
We are interested in the problem of intelligent connection of perception to action, i.e., the con... more We are interested in the problem of intelligent connection of perception to action, i.e., the connection between numerical data and cognitive functions. In this paper we extend conventional quantity space into that in a geometric vector context and then propose quantity arithmetic for quantity vector computation in a normalized quantity space. An example of motion abstraction of a Puma robot is provided to demonstrate the effectiveness of the proposed method.
Classification of human motion in video data is essential in numerous applications. However, prob... more Classification of human motion in video data is essential in numerous applications. However, problems arise as the human exhibits complex and dynamic motion that is nonlinear and time varying. In this paper, we propose a knowledge-based human motion classification framework that employs fuzzy qualitative reasoning to address these problems. Our approach utilises the rich contextual information (e.g. structural and transitional characteristic of human motion) captured in video sequence to effectively study and recognise human motion. With the aid of domain knowledge, a set of fuzzy rules are defined in the knowledge base. This work is in contrast with previous attempts that depend solely on the trajectories of the body parts. Experimental results on two classes of motion (e.g. walking and running) that result in similar motions; and a comparison with the conventional method has demonstrated and validated the effectiveness of the proposed method in improving the perception of human motion.
We propose a novel method to identify the correspondence of objects in different spaces using pan... more We propose a novel method to identify the correspondence of objects in different spaces using panorama generation and qualitative reasoning, the spaces are namely image space, camera fuzzy qualitative space and real world space. The correspondence is carried out in a three-layered image understanding framework. The first layer consists of single cameras which is to extract quantitative meausres using off-the-shelf image algorithms with a target of providing local feature information; The second layer targets at fusing qualitative information of single cameras at the level of cameras network; The third layer is intended to generate semantic description of object behaviours using nature language generation. This paper is focused on qualitative correspondence of objects in the first layer, which is realized by a two-stage tracking cycle consisting of panorama generation and object tracking. A case study is given to demonstrate the effectiveness of the method.
The viewpoint issue has been one of the bottlenecks for research development and practical implem... more The viewpoint issue has been one of the bottlenecks for research development and practical implementation of human motion analysis. In this paper, we introduce a new method, e.g., hidden conditional random fields(HCRFs) to achieve viewpoint insensitive human action recognition. The HCRF model can relax the independence assumption of the generative models. So it is very suitable to model the human actions from different actors and different viewpoints. Experiment results on a public dataset demonstrate the effectiveness and robustness of our method.
... 613 Ahmad Ghanbari, Mir Masoud Seyyed Fakhrabadi, and Ali Rostami Optimum ... 43 Ercan, M. Fi... more ... 613 Ahmad Ghanbari, Mir Masoud Seyyed Fakhrabadi, and Ali Rostami Optimum ... 43 Ercan, M. Fikret 744 Fakhrabadi, Mir Masoud Seyyed 613 Fan, Baojie 876, 929 Fan, Changchun 802 Fan, Shaoshuai 687 Feng, Hongpeng
As a temporal classification problem, visual-based human actions recognition is an important comp... more As a temporal classification problem, visual-based human actions recognition is an important component for some potential applications. In this paper, we combine Fuzzy Principle Component Analysis(Fuzzy PCA) and hidden Conditional Random Fields(HCRFs) to achieve a viewpoint insensitive human action recognition. Fuzzy PCA is used to reduce the dimension of the silhouette image features to obtain the compact representation of action space. HCRFs is applied to model the human actions from different actors and different viewpoints. This method can relax the independence assumption of the generative model. Experiment results on a public dataset demonstrate the effectiveness and robustness of our method.
Journal of Advanced Computational Intelligence and Intelligent Informatics, 2010
... Furthermore, we use iPhone and iPod touch for cognitive rehabilitation therapy. ... [2] S.Att... more ... Furthermore, we use iPhone and iPod touch for cognitive rehabilitation therapy. ... [2] S.Attygalle, M.Duff, T.Rikakis, Jiping.H, Low-cost, at-home assessment system with Wii Remote based motion capture, In Proceedings of the Conference on Virtual Rehabilitation, pp.168-174 ...
Automatic human action recognition has been a challenging issue in the field of machine vision. S... more Automatic human action recognition has been a challenging issue in the field of machine vision. Some high-level features such as SIFT, although with promising performance for action recognition, are computationally complex to some extent. To deal with this problem, we construct the features based on the Distance Transform of body contours, which is relatively simple and computationally efficient, to represent human action in the video. After extracting the features from videos, we adopt the Conditional Random Field for modeling the temporal action sequences. The proposed method is tested with an available standard dataset. We also testify the robustness of our method on various realistic conditions, such as body occlusion or intersection.
How to recognize human action from videos captured by modern cameras efficiently and effectively ... more How to recognize human action from videos captured by modern cameras efficiently and effectively is a challenge in real applications. Traditional methods which need professional analysts are facing a bottleneck because of their shortcomings. To cope with the disadvantage, methods based on computer vision techniques, without or with only a few human interventions, have been proposed to analyse human actions in videos automatically. This paper provides a method combining the three dimensional Scale Invariant Feature Transform (SIFT) detector and the Latent Dirichlet Allocation (LDA) model for human motion analysis. To represent videos effectively and robustly, we extract the 3D SIFT descriptor around each interest point, which is sampled densely from 3D Space-time video volumes. After obtaining the representation of each video frame, the LDA model is adopted to discover the underlying structure-the categorization of human actions in the collection of videos. Public available standard datasets are used to test our method. The concluding part discusses the research challenges and future directions.
... Rectilinear Snake Robot..... 613 Ahmad Ghanbari, Mir Masoud SeyyedFakhrabadi, and Ali Rostami... more ... Rectilinear Snake Robot..... 613 Ahmad Ghanbari, Mir Masoud SeyyedFakhrabadi, and Ali Rostami Optimum Dynamic Modeling of a Wall Climbing Robot for Ship Rust Removal..... 623 Xingru Wang ...
... Asian Jornal of Control , 12: 704713. [CrossRef] View all references), state estimation (Bal... more ... Asian Jornal of Control , 12: 704713. [CrossRef] View all references), state estimation (Balasubramaniam, Lakshmanan, and Jeeva Sathya Theesar 20101. Balasubramaniam, P, Lakshmanan, S and Jeeva Sathya Theesar, S. 2010. ... Yu, JJ, Zhang, KJ and Fei, SM. 2009. ...
... Cao, Jiangtao, Li, P. and Liu, Honghai (2009) Improved interval type-2 fuzzy logic controller... more ... Cao, Jiangtao, Li, P. and Liu, Honghai (2009) Improved interval type-2 fuzzy logic controller (Chinese). Control and Decision, 24 (10). ... Departments/Research Groups: Faculty of Creative and Cultural Industries > School of Creative Technologies. Depositing User: Trish Skinner. ...
An exemplar-based view-invariant human action recognition framework is proposed to recognize the ... more An exemplar-based view-invariant human action recognition framework is proposed to recognize the human actions from any arbitrary viewpoint image sequence. In this framework, human action is modelled as a sequence of body key poses (i.e., exemplars) which are represented by a collection of silhouette images. The human actions are recognized by matching the observed image sequence to predefined exemplars, in which the temporal constraints are imposed in the exemplar-based Hidden Markov Model (HMM). Furthermore, a new two-level recognition framework is introduced to improve the discrimination capability for the similar human actions. The aim of the first level recognition is to decide an equivalent set in which the testing action is included instead of directly achieving the final recognition results. In the second level, the weighted contour shape feature is used to calculate the observation probability to discriminate the similar actions. The proposed framework is evaluated in a public dataset and the results show that it not only reduces computational complexity, but it is also able to accurately recognize human actions using single cameras. Besides it is verified that the weighted contour shape feature is effective to differentiate the similar arm-related actions.
We are interested in the problem of intelligent connection of perception to action, i.e., the con... more We are interested in the problem of intelligent connection of perception to action, i.e., the connection between numerical data and cognitive functions. In this paper we extend conventional quantity space into that in a geometric vector context and then propose quantity arithmetic for quantity vector computation in a normalized quantity space. An example of motion abstraction of a Puma robot is provided to demonstrate the effectiveness of the proposed method.
Classification of human motion in video data is essential in numerous applications. However, prob... more Classification of human motion in video data is essential in numerous applications. However, problems arise as the human exhibits complex and dynamic motion that is nonlinear and time varying. In this paper, we propose a knowledge-based human motion classification framework that employs fuzzy qualitative reasoning to address these problems. Our approach utilises the rich contextual information (e.g. structural and transitional characteristic of human motion) captured in video sequence to effectively study and recognise human motion. With the aid of domain knowledge, a set of fuzzy rules are defined in the knowledge base. This work is in contrast with previous attempts that depend solely on the trajectories of the body parts. Experimental results on two classes of motion (e.g. walking and running) that result in similar motions; and a comparison with the conventional method has demonstrated and validated the effectiveness of the proposed method in improving the perception of human motion.
We propose a novel method to identify the correspondence of objects in different spaces using pan... more We propose a novel method to identify the correspondence of objects in different spaces using panorama generation and qualitative reasoning, the spaces are namely image space, camera fuzzy qualitative space and real world space. The correspondence is carried out in a three-layered image understanding framework. The first layer consists of single cameras which is to extract quantitative meausres using off-the-shelf image algorithms with a target of providing local feature information; The second layer targets at fusing qualitative information of single cameras at the level of cameras network; The third layer is intended to generate semantic description of object behaviours using nature language generation. This paper is focused on qualitative correspondence of objects in the first layer, which is realized by a two-stage tracking cycle consisting of panorama generation and object tracking. A case study is given to demonstrate the effectiveness of the method.
The viewpoint issue has been one of the bottlenecks for research development and practical implem... more The viewpoint issue has been one of the bottlenecks for research development and practical implementation of human motion analysis. In this paper, we introduce a new method, e.g., hidden conditional random fields(HCRFs) to achieve viewpoint insensitive human action recognition. The HCRF model can relax the independence assumption of the generative models. So it is very suitable to model the human actions from different actors and different viewpoints. Experiment results on a public dataset demonstrate the effectiveness and robustness of our method.
... 613 Ahmad Ghanbari, Mir Masoud Seyyed Fakhrabadi, and Ali Rostami Optimum ... 43 Ercan, M. Fi... more ... 613 Ahmad Ghanbari, Mir Masoud Seyyed Fakhrabadi, and Ali Rostami Optimum ... 43 Ercan, M. Fikret 744 Fakhrabadi, Mir Masoud Seyyed 613 Fan, Baojie 876, 929 Fan, Changchun 802 Fan, Shaoshuai 687 Feng, Hongpeng
As a temporal classification problem, visual-based human actions recognition is an important comp... more As a temporal classification problem, visual-based human actions recognition is an important component for some potential applications. In this paper, we combine Fuzzy Principle Component Analysis(Fuzzy PCA) and hidden Conditional Random Fields(HCRFs) to achieve a viewpoint insensitive human action recognition. Fuzzy PCA is used to reduce the dimension of the silhouette image features to obtain the compact representation of action space. HCRFs is applied to model the human actions from different actors and different viewpoints. This method can relax the independence assumption of the generative model. Experiment results on a public dataset demonstrate the effectiveness and robustness of our method.
Journal of Advanced Computational Intelligence and Intelligent Informatics, 2010
... Furthermore, we use iPhone and iPod touch for cognitive rehabilitation therapy. ... [2] S.Att... more ... Furthermore, we use iPhone and iPod touch for cognitive rehabilitation therapy. ... [2] S.Attygalle, M.Duff, T.Rikakis, Jiping.H, Low-cost, at-home assessment system with Wii Remote based motion capture, In Proceedings of the Conference on Virtual Rehabilitation, pp.168-174 ...
Automatic human action recognition has been a challenging issue in the field of machine vision. S... more Automatic human action recognition has been a challenging issue in the field of machine vision. Some high-level features such as SIFT, although with promising performance for action recognition, are computationally complex to some extent. To deal with this problem, we construct the features based on the Distance Transform of body contours, which is relatively simple and computationally efficient, to represent human action in the video. After extracting the features from videos, we adopt the Conditional Random Field for modeling the temporal action sequences. The proposed method is tested with an available standard dataset. We also testify the robustness of our method on various realistic conditions, such as body occlusion or intersection.
How to recognize human action from videos captured by modern cameras efficiently and effectively ... more How to recognize human action from videos captured by modern cameras efficiently and effectively is a challenge in real applications. Traditional methods which need professional analysts are facing a bottleneck because of their shortcomings. To cope with the disadvantage, methods based on computer vision techniques, without or with only a few human interventions, have been proposed to analyse human actions in videos automatically. This paper provides a method combining the three dimensional Scale Invariant Feature Transform (SIFT) detector and the Latent Dirichlet Allocation (LDA) model for human motion analysis. To represent videos effectively and robustly, we extract the 3D SIFT descriptor around each interest point, which is sampled densely from 3D Space-time video volumes. After obtaining the representation of each video frame, the LDA model is adopted to discover the underlying structure-the categorization of human actions in the collection of videos. Public available standard datasets are used to test our method. The concluding part discusses the research challenges and future directions.
... Rectilinear Snake Robot..... 613 Ahmad Ghanbari, Mir Masoud SeyyedFakhrabadi, and Ali Rostami... more ... Rectilinear Snake Robot..... 613 Ahmad Ghanbari, Mir Masoud SeyyedFakhrabadi, and Ali Rostami Optimum Dynamic Modeling of a Wall Climbing Robot for Ship Rust Removal..... 623 Xingru Wang ...
Uploads
Papers by Honghai Liu