Volume 11, Number 5, 2014
Special Issue on Massive Visual Computing (pp.459-516)
column
Display Method:
2014, vol. 11, no. 5, pp. 459-468, doi: 10.1007/s11633-014-0835-0
Abstract:
In expression recognition, feature representation is critical for successful recognition since it contains distinctive information of expressions. In this paper, a new approach for representing facial expression features is proposed with its objective to describe features in an effective and efficient way in order to improve the recognition performance. The method combines the facial action coding system (FACS) and "uniform" local binary patterns (LBP) to represent facial expression features from coarse to fine. The facial feature regions are extracted by active shape models (ASM) based on FACS to obtain the gray-level texture. Then, LBP is used to represent expression features for enhancing the discriminant. A facial expression recognition system is developed based on this feature extraction method by using K nearest neighborhood (K-NN) classifier to recognize facial expressions. Finally, experiments are carried out to evaluate this feature extraction method. The significance of removing the unrelated facial regions and enhancing the discrimination ability of expression features in the recognition process is indicated by the results, in addition to its convenience.
In expression recognition, feature representation is critical for successful recognition since it contains distinctive information of expressions. In this paper, a new approach for representing facial expression features is proposed with its objective to describe features in an effective and efficient way in order to improve the recognition performance. The method combines the facial action coding system (FACS) and "uniform" local binary patterns (LBP) to represent facial expression features from coarse to fine. The facial feature regions are extracted by active shape models (ASM) based on FACS to obtain the gray-level texture. Then, LBP is used to represent expression features for enhancing the discriminant. A facial expression recognition system is developed based on this feature extraction method by using K nearest neighborhood (K-NN) classifier to recognize facial expressions. Finally, experiments are carried out to evaluate this feature extraction method. The significance of removing the unrelated facial regions and enhancing the discrimination ability of expression features in the recognition process is indicated by the results, in addition to its convenience.
2014, vol. 11, no. 5, pp. 469-479, doi: 10.1007/s11633-014-0829-y
Abstract:
This paper puts forward a method for abdomen panorama reconstruction based on a stereo vision system. For the purpose of recovering the abdomen completely and accurately under the condition of actual photographing with illumination variance and blur noise, some innovative combined feature descriptors are presented on the basis of Hu-moment invariants. Furthermore, considering the study on the abdomen surface reconstruction, a circle template which is divided into 6 sectors is designed. It is noted that a descriptor merely using gray intensity is not able to provide sufficient information for feature description. Consequently, the sector entropy which denotes the structure characteristics is drawn into the feature descriptor. By means of the combined effect of the gray intensity and the sector entropy, the similarity measurement is conducted for the final abdomen reconstruction. The experimental results reveal that the proposed method can acquire a high precision of abdomen reconstruction similar to the 3D scanner. This stereo vision system has wide practicability in the field of clothing.
This paper puts forward a method for abdomen panorama reconstruction based on a stereo vision system. For the purpose of recovering the abdomen completely and accurately under the condition of actual photographing with illumination variance and blur noise, some innovative combined feature descriptors are presented on the basis of Hu-moment invariants. Furthermore, considering the study on the abdomen surface reconstruction, a circle template which is divided into 6 sectors is designed. It is noted that a descriptor merely using gray intensity is not able to provide sufficient information for feature description. Consequently, the sector entropy which denotes the structure characteristics is drawn into the feature descriptor. By means of the combined effect of the gray intensity and the sector entropy, the similarity measurement is conducted for the final abdomen reconstruction. The experimental results reveal that the proposed method can acquire a high precision of abdomen reconstruction similar to the 3D scanner. This stereo vision system has wide practicability in the field of clothing.
2014, vol. 11, no. 5, pp. 480-488, doi: 10.1007/s11633-014-0833-2
Abstract:
This paper proposes a new two-phase approach to robust text detection by integrating the visual appearance and the geometric reasoning rules. In the first phase, geometric rules are used to achieve a higher recall rate. Specifically, a robust stroke width transform (RSWT) feature is proposed to better recover the stroke width by additionally considering the cross of two strokes and the continuousness of the letter border. In the second phase, a classification scheme based on visual appearance features is used to reject the false alarms while keeping the recall rate. To learn a better classifier from multiple visual appearance features, a novel classification method called double soft multiple kernel learning (DS-MKL) is proposed. DS-MKL is motivated by a novel kernel margin perspective for multiple kernel learning and can effectively suppress the influence of noisy base kernels. Comprehensive experiments on the benchmark ICDAR2005 competition dataset demonstrate the effectiveness of the proposed two-phase text detection approach over the state-of-the-art approaches by a performance gain up to 4.4% in terms of F-measure.
This paper proposes a new two-phase approach to robust text detection by integrating the visual appearance and the geometric reasoning rules. In the first phase, geometric rules are used to achieve a higher recall rate. Specifically, a robust stroke width transform (RSWT) feature is proposed to better recover the stroke width by additionally considering the cross of two strokes and the continuousness of the letter border. In the second phase, a classification scheme based on visual appearance features is used to reject the false alarms while keeping the recall rate. To learn a better classifier from multiple visual appearance features, a novel classification method called double soft multiple kernel learning (DS-MKL) is proposed. DS-MKL is motivated by a novel kernel margin perspective for multiple kernel learning and can effectively suppress the influence of noisy base kernels. Comprehensive experiments on the benchmark ICDAR2005 competition dataset demonstrate the effectiveness of the proposed two-phase text detection approach over the state-of-the-art approaches by a performance gain up to 4.4% in terms of F-measure.
2014, vol. 11, no. 5, pp. 489-499, doi: 10.1007/s11633-014-0832-3
Abstract:
Multimedia is one of the important communication channels for mankind. Due to the advancement in technology and enormous growth of mankind, a vast array of multimedia data is available today. This has resulted in the obvious need for some techniques for retrieving these data. This paper will give an overview of ontology-based image retrieval system for asteroideae flower family domain. In order to reduce the semantic gap between the low-level visual features of an image and the high-level domain knowledge, we have incorporated a concept of multi-modal image ontology. So, the created asteroideae flower domain specific ontology would have the knowledge about the domain and the visual features. The visual features used to define the ontology are prevalent color, basic intrinsic pattern and contour gradient. In prevalent color extraction, the most dominant color from the images was identified and indexed. In order to determine the texture pattern for a particular flower, basic intrinsic patterns were used. The contour gradients provide the information on the image edges with respect to the image base. These feature values are embedded in the ontology at appropriate slots with respect to the domain knowledge. This paper also defines some of the query axioms which are used to retrieve appropriate information from the created ontology. This ontology can be used for image retrieval system in semantic web.
Multimedia is one of the important communication channels for mankind. Due to the advancement in technology and enormous growth of mankind, a vast array of multimedia data is available today. This has resulted in the obvious need for some techniques for retrieving these data. This paper will give an overview of ontology-based image retrieval system for asteroideae flower family domain. In order to reduce the semantic gap between the low-level visual features of an image and the high-level domain knowledge, we have incorporated a concept of multi-modal image ontology. So, the created asteroideae flower domain specific ontology would have the knowledge about the domain and the visual features. The visual features used to define the ontology are prevalent color, basic intrinsic pattern and contour gradient. In prevalent color extraction, the most dominant color from the images was identified and indexed. In order to determine the texture pattern for a particular flower, basic intrinsic patterns were used. The contour gradients provide the information on the image edges with respect to the image base. These feature values are embedded in the ontology at appropriate slots with respect to the domain knowledge. This paper also defines some of the query axioms which are used to retrieve appropriate information from the created ontology. This ontology can be used for image retrieval system in semantic web.
2014, vol. 11, no. 5, pp. 500-509, doi: 10.1007/s11633-014-0831-4
Abstract:
Most of the exist action recognition methods mainly utilize spatio-temporal descriptors of single interest point while ignoring their potential integral information, such as spatial distribution information. By combining local spatio-temporal feature and global positional distribution information (PDI) of interest points, a novel motion descriptor is proposed in this paper. The proposed method detects interest points by using an improved interest point detection method. Then, 3-dimensional scale-invariant feature transform (3D SIFT) descriptors are extracted for every interest point. In order to obtain a compact description and efficient computation, the principal component analysis (PCA) method is utilized twice on the 3D SIFT descriptors of single frame and multiple frames. Simultaneously, the PDI of the interest points are computed and combined with the above features. The combined features are quantified and selected and finally tested by using the support vector machine (SVM) recognition algorithm on the public KTH dataset. The testing results have showed that the recognition rate has been significantly improved and the proposed features can more accurately describe human motion with high adaptability to scenarios.
Most of the exist action recognition methods mainly utilize spatio-temporal descriptors of single interest point while ignoring their potential integral information, such as spatial distribution information. By combining local spatio-temporal feature and global positional distribution information (PDI) of interest points, a novel motion descriptor is proposed in this paper. The proposed method detects interest points by using an improved interest point detection method. Then, 3-dimensional scale-invariant feature transform (3D SIFT) descriptors are extracted for every interest point. In order to obtain a compact description and efficient computation, the principal component analysis (PCA) method is utilized twice on the 3D SIFT descriptors of single frame and multiple frames. Simultaneously, the PDI of the interest points are computed and combined with the above features. The combined features are quantified and selected and finally tested by using the support vector machine (SVM) recognition algorithm on the public KTH dataset. The testing results have showed that the recognition rate has been significantly improved and the proposed features can more accurately describe human motion with high adaptability to scenarios.
2014, vol. 11, no. 5, pp. 510-516, doi: 10.1007/s11633-014-0830-5
Abstract:
Design of video encoders involves implementation of fast mode decision (FMD) algorithm to reduce computation complexity while maintaining the performance of the coding. Although H.264/scalable video coding (SVC) achieves high scalability and coding efficiency, it also has high complexity in implementing its exhaustive computation. In this paper, a novel algorithm is proposed to reduce the redundant candidate modes by making use of the correlation among layers. A desired mode list is created based on the probability to be the best mode for each block in base layer and a candidate mode selection in the enhancement layer by the correlations of modes among reference frame and current frame. Our algorithm is implemented in joint scalable video model (JSVM) 9.19.15 reference software and the performance is evaluated based on the average encoding time, peak signal to noise ration (PSNR) and bit rate. The experimental results show 41.89% improvement in encoding time with minimal loss of 0.02 dB in PSNR and 0.05% increase in bit rate.
Design of video encoders involves implementation of fast mode decision (FMD) algorithm to reduce computation complexity while maintaining the performance of the coding. Although H.264/scalable video coding (SVC) achieves high scalability and coding efficiency, it also has high complexity in implementing its exhaustive computation. In this paper, a novel algorithm is proposed to reduce the redundant candidate modes by making use of the correlation among layers. A desired mode list is created based on the probability to be the best mode for each block in base layer and a candidate mode selection in the enhancement layer by the correlations of modes among reference frame and current frame. Our algorithm is implemented in joint scalable video model (JSVM) 9.19.15 reference software and the performance is evaluated based on the average encoding time, peak signal to noise ration (PSNR) and bit rate. The experimental results show 41.89% improvement in encoding time with minimal loss of 0.02 dB in PSNR and 0.05% increase in bit rate.
2014, vol. 11, no. 5, pp. 517-526, doi: 10.1007/s11633-014-0834-1
Abstract:
Nowadays, high-precision motion controls are needed in modern manufacturing industry. A data-driven nonparametric model adaptive control (NMAC) method is proposed in this paper to control the position of a linear servo system. The controller design requires no information about the structure of linear servo system, and it is based on the estimation and forecasting of the pseudo-partial derivatives (PPD) which are estimated according to the voltage input and position output of the linear motor. The characteristics and operational mechanism of the permanent magnet synchronous linear motor (PMSLM) are introduced, and the proposed nonparametric model control strategy has been compared with the classic proportional-integral-derivative (PID) control algorithm. Several real-time experiments on the motion control system incorporating a permanent magnet synchronous linear motor showed that the nonparametric model adaptive control method improved the system s response to disturbances and its position-tracking precision, even for a nonlinear system with incompletely known dynamic characteristics.
Nowadays, high-precision motion controls are needed in modern manufacturing industry. A data-driven nonparametric model adaptive control (NMAC) method is proposed in this paper to control the position of a linear servo system. The controller design requires no information about the structure of linear servo system, and it is based on the estimation and forecasting of the pseudo-partial derivatives (PPD) which are estimated according to the voltage input and position output of the linear motor. The characteristics and operational mechanism of the permanent magnet synchronous linear motor (PMSLM) are introduced, and the proposed nonparametric model control strategy has been compared with the classic proportional-integral-derivative (PID) control algorithm. Several real-time experiments on the motion control system incorporating a permanent magnet synchronous linear motor showed that the nonparametric model adaptive control method improved the system s response to disturbances and its position-tracking precision, even for a nonlinear system with incompletely known dynamic characteristics.
2014, vol. 11, no. 5, pp. 527-535, doi: 10.1007/s11633-014-0821-6
Abstract:
We present an adaptive control scheme of accumulative error to stabilize the unstable fixed point for chaotic systems which only satisfies local Lipschitz condition, and discuss how the convergence factor affects the convergence and the characteristics of the final control strength. We define a minimal local Lipschitz coefficient, which can enlarge the condition of chaos control. Compared with other adaptive methods, this control scheme is simple and easy to implement by integral circuits in practice. It is also robust against the effect of noise. These are illustrated with numerical examples.
We present an adaptive control scheme of accumulative error to stabilize the unstable fixed point for chaotic systems which only satisfies local Lipschitz condition, and discuss how the convergence factor affects the convergence and the characteristics of the final control strength. We define a minimal local Lipschitz coefficient, which can enlarge the condition of chaos control. Compared with other adaptive methods, this control scheme is simple and easy to implement by integral circuits in practice. It is also robust against the effect of noise. These are illustrated with numerical examples.
2014, vol. 11, no. 5, pp. 536-542, doi: 10.1007/s11633-014-0828-z
Abstract:
The power output of the photovoltaic (PV) system having multiple arrays gets reduced to a great extent when it is partially shaded due to environmental hindrances. The maximum power trackers which are conventionally used may not be competent enough to find the maximum power point (MPP) during partially shaded conditions. The sensible reason for the failure of conventional trackers is during partial shaded conditions the PV arrays exhibit multi peak power curves, thereby making simple maximum power point tracking (MPPT) algorithms like perturb and observe (P&O) to get stuck with local maxima instead of capturing global maxima. Therefore, global search MPPT aided by evolutionary and swarm intelligence algorithms will be conducive to find global power point during partially shaded conditions. This work suggests a unified controller which feeds control signal to its power electronic conditioner placed at each module. The evolutionary algorithm which is taken into consideration in this work is differential evolution (DE). The performance of the proposed method is compared to the classical un-dimensional search controller and it is evident from the Matlab/Simulink results that the unified controller prevails over the distributed counterpart.
The power output of the photovoltaic (PV) system having multiple arrays gets reduced to a great extent when it is partially shaded due to environmental hindrances. The maximum power trackers which are conventionally used may not be competent enough to find the maximum power point (MPP) during partially shaded conditions. The sensible reason for the failure of conventional trackers is during partial shaded conditions the PV arrays exhibit multi peak power curves, thereby making simple maximum power point tracking (MPPT) algorithms like perturb and observe (P&O) to get stuck with local maxima instead of capturing global maxima. Therefore, global search MPPT aided by evolutionary and swarm intelligence algorithms will be conducive to find global power point during partially shaded conditions. This work suggests a unified controller which feeds control signal to its power electronic conditioner placed at each module. The evolutionary algorithm which is taken into consideration in this work is differential evolution (DE). The performance of the proposed method is compared to the classical un-dimensional search controller and it is evident from the Matlab/Simulink results that the unified controller prevails over the distributed counterpart.
2014, vol. 11, no. 5, pp. 543-554, doi: 10.1007/s11633-014-0827-0
Abstract:
The novel eye-based human-computer interaction (HCI) system aims to provide people, especially, disabled persons, a new way of communication with surroundings. It adopts a series of continual eye movements as input to perform simple control activities. Identification of eye movements is the crucial technology in these eye-based HCI systems. At present, researches on eye movement identification mainly focus on frontal face images. In fact, acquisition of non-frontal face images is more reasonable in real applications. In this paper, we discuss the identification process of eye movements from non-frontal face images. Firstly, the original head-shoulder images of 0°-±60° azimuths are sampled without any auxiliary light source. Secondly, the non-frontal face region is detected by using the Adaboost cascade classifiers. After that, we roughly extract eye windows by the integral projection function. Then, we propose a new method to calculate the x−y coordinates of the pupil center point by searching the minimal intensity value in the eye windows. According to the trajectory of the pupil center points, different eye movements (eye moving left, right, up or down) are successfully identified. A set of experiments is presented.
The novel eye-based human-computer interaction (HCI) system aims to provide people, especially, disabled persons, a new way of communication with surroundings. It adopts a series of continual eye movements as input to perform simple control activities. Identification of eye movements is the crucial technology in these eye-based HCI systems. At present, researches on eye movement identification mainly focus on frontal face images. In fact, acquisition of non-frontal face images is more reasonable in real applications. In this paper, we discuss the identification process of eye movements from non-frontal face images. Firstly, the original head-shoulder images of 0°-±60° azimuths are sampled without any auxiliary light source. Secondly, the non-frontal face region is detected by using the Adaboost cascade classifiers. After that, we roughly extract eye windows by the integral projection function. Then, we propose a new method to calculate the x−y coordinates of the pupil center point by searching the minimal intensity value in the eye windows. According to the trajectory of the pupil center points, different eye movements (eye moving left, right, up or down) are successfully identified. A set of experiments is presented.
2014, vol. 11, no. 5, pp. 555-564, doi: 10.1007/s11633-014-0826-1
Abstract:
This paper presents a bio-inspired backstepping adaptive sliding mode control strategy for a novel 3 degree of freedom (3-DOF) parallel mechanism with actuation redundancy. Based on the kinematic model and the dynamic model, a sliding mode controller is designed to assure the tracking performance, and an adaptive law is introduced to approximate the system uncertainty including parameters variation, external disturbances and un-modeled part. Furthermore, a bio-inspired model is introduced to solve the inherent chattering problem of sliding mode control and provide a chattering free control. The simulation and experimental results testify that the proposed bio-inspired backstepping adaptive sliding mode control can achieve better performance (the tracking accuracy, robustness, response speed, etc.) than the conventional slide mode control.
This paper presents a bio-inspired backstepping adaptive sliding mode control strategy for a novel 3 degree of freedom (3-DOF) parallel mechanism with actuation redundancy. Based on the kinematic model and the dynamic model, a sliding mode controller is designed to assure the tracking performance, and an adaptive law is introduced to approximate the system uncertainty including parameters variation, external disturbances and un-modeled part. Furthermore, a bio-inspired model is introduced to solve the inherent chattering problem of sliding mode control and provide a chattering free control. The simulation and experimental results testify that the proposed bio-inspired backstepping adaptive sliding mode control can achieve better performance (the tracking accuracy, robustness, response speed, etc.) than the conventional slide mode control.
2014, vol. 11, no. 5, pp. 565-570, doi: 10.1007/s11633-014-0813-6
Abstract:
In this paper, a fire-new general integral control, named general convex integral control, is proposed. It is derived by defining a nonlinear function set to form the integral control action and educe a new convex function gain integrator, introducing the partial derivative of Lyapunov function into the integrator and resorting to a general strategy to transform ordinary control into general integral control. By using Lyapunov method along with the LaSalle s invariance principle, the theorem to ensure regionally as well as semi-globally asymptotic stability is established only by some bounded information. Moreover, the lemma to ensure the integrator output to be bounded in the time domain is proposed. The highlight point of this integral control strategy is that the integral control action seems to be infinity, but it factually is finite in the time domain. Therefore, a simple and ingenious method to design the general integral control is founded. Simulation results showed that under the normal and perturbed cases, the optimum response in the whole control domain of interest can all be achieved by a set of control gains, even under the case that the payload is changed abruptly.
In this paper, a fire-new general integral control, named general convex integral control, is proposed. It is derived by defining a nonlinear function set to form the integral control action and educe a new convex function gain integrator, introducing the partial derivative of Lyapunov function into the integrator and resorting to a general strategy to transform ordinary control into general integral control. By using Lyapunov method along with the LaSalle s invariance principle, the theorem to ensure regionally as well as semi-globally asymptotic stability is established only by some bounded information. Moreover, the lemma to ensure the integrator output to be bounded in the time domain is proposed. The highlight point of this integral control strategy is that the integral control action seems to be infinity, but it factually is finite in the time domain. Therefore, a simple and ingenious method to design the general integral control is founded. Simulation results showed that under the normal and perturbed cases, the optimum response in the whole control domain of interest can all be achieved by a set of control gains, even under the case that the payload is changed abruptly.
_00.jpg)
