Papers - THI THI ZIN
-
A data-driven key information search system in big data analytics
Tin P., Zin T., Hama H., Toriu T.
ICIC Express Letters, Part B: Applications 5 ( 2 ) 365 - 370 2014.2
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters, Part B: Applications
Nowadays, the emergence of data-driven information trends has exploded the volume, velocity and variety referring to as Big Data. Analyzing and understanding the Big Data for key information trends and unusually high or low activity is an important problem in the Big Data analytics having many applications. In this paper, we propose a framework for a data-driven key information search system in Big Data analytics. Since many of the Big Data problems are dependent on the unstructured data analysis, the proposed framework is based on an arbitrarily related data records and metric spaces to study such problems. In doing so, a set theoretic data structure is employed for expressing non-intrinsically related massive large data sets and transforming into a stochastic model to compute a new stream of right information for right targets at right time. By processing a steady state distribution of the model based on real-time data, the framework enables to make time-sensitive decisions, monitor emerging trends, and jump on new insight information. As an illustration, the unstructured data targeted in this work to organize, is the public consuming patterns available in social network platforms and in a kind of various types of sensor readings, followed by time series data analysis to obtain the necessary key information over Big Data. © 2014 ISSN 2185-2766.
-
An embedded knowledge extraction technology for consumer video surveillance Reviewed
Thi Thi Zin, Pyke Tin, T. Toriu, H. Hama
2014 IEEE 3rd Global Conference on Consumer Electronics, GCCE 2014 510 - 511 2014.2
Language：English Publishing type：Research paper (international conference proceedings) Publisher：2014 IEEE 3rd Global Conference on Consumer Electronics, GCCE 2014
New advances in embedded computing technology have opened up the potential for new era of consumer surveillance systems. This paper will explore and propose a new embedded modeling technique for the configuration of consumer video surveillance systems that can identify events of interest, especially on abandoned and stolen objects in indoor and outdoor environments. The proposed embedded system will focus on high level behavior understanding for object detection, tracking and classification. The experimental results illustrate the ability of the system to create complex spatiotemporal relations and to recognize the behavior of one or multiple objects in various video scenes.
-
An integrated framework for detecting suspicious behaviors in video surveillance
Zin T., Tin P., Hama H., Toriu T.
Proceedings of SPIE - The International Society for Optical Engineering 9026 2014
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Proceedings of SPIE - The International Society for Optical Engineering
In this paper, we propose an integrated framework for detecting suspicious behaviors in video surveillance systems which are established in public places such as railway stations, airports, shopping malls and etc. Especially, people loitering in suspicion, unattended objects left behind and exchanging suspicious objects between persons are common security concerns in airports and other transit scenarios. These involve understanding scene/event, analyzing human movements, recognizing controllable objects, and observing the effect of the human movement on those objects. In the proposed framework, multiple background modeling technique, high level motion feature extraction method and embedded Markov chain models are integrated for detecting suspicious behaviors in real time video surveillance systems. Specifically, the proposed framework employs probability based multiple backgrounds modeling technique to detect moving objects. Then the velocity and distance measures are computed as the high level motion features of the interests. By using an integration of the computed features and the first passage time probabilities of the embedded Markov chain, the suspicious behaviors in video surveillance are analyzed for detecting loitering persons, objects left behind and human interactions such as fighting. The proposed framework has been tested by using standard public datasets and our own video surveillance scenarios. © 2014 SPIE.
DOI： 10.1117/12.2041232
-
New information search system in social networks application to disaster event analysis
Zin T., Tin P., Toriu T., Hama H.
IAENG Transactions on Engineering Sciences - Special Issue of the International MultiConference of Engineers and Computer Scientists, IMECS 2013 and World Congress on Engineering, WCE 2013 411 - 419 2014
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：IAENG Transactions on Engineering Sciences - Special Issue of the International MultiConference of Engineers and Computer Scientists, IMECS 2013 and World Congress on Engineering, WCE 2013
This chapter is concerned with the development of a new information search system in social network platforms for analyzing events in case of natural and man-made disasters. In todays, the world has witnessed the role of many social networks such as Facebook, Flicker, Twitter, and YouTube for extracting key and crucial information when worldwide scale major events had occurred. A social network can also be considered as information link creator between users and available resources. This role has been boosted in tremendous amount during and after the recent disasters around the world; for example, the 2010 Philippine typhoon, the 2010 Haiti earthquake, the 2011 Brazil flood, and the 2011 Japan earthquake and tsunami and Boston Marathon Explosion 2013. In such emergency situations, extracting and analyzing key information from the congested information resources posted or hosted in variety of forms and norms through almost all social network platforms are of major concerns in assessing the situation and in decision making. Therefore, some mathematical modeling techniques of branching processes and Markov chain theory to investigate how new knowledge and new information about the disasters spreads on the social networks and how to extract trust and reliable key and dominant information are discussed here. Specifically, a set of text messages and visual information is transformed into a Markov chain to produce stationary and time dependent distributions. Then, the abnormalities and suspicious patterns occurring in the distributions are analyzed for detecting and extracting key information. Finally, some simulation results are given in the case of Boston Marathon Explosion occurred in April 15, 2013 by using three social networks Flickr, YouTube and Twitter information. © 2014 Taylor & Francis Group, London.
-
A Big Data application framework for consumer behavior analysis Reviewed
Thi Thi Zin, Pyke Tin, T. Toriu, H. Hama
2013 IEEE 2nd Global Conference on Consumer Electronics, GCCE 2013 245 - 246 2013.12
Language：English Publishing type：Research paper (international conference proceedings) Publisher：2013 IEEE 2nd Global Conference on Consumer Electronics, GCCE 2013
More than ever before, the amount of data about consumers, suppliers and products has been exploding in today consumer world referred as 'Big Data'. In addition, more data is available to the consumer world from multiple sources including social network platforms. In order to deal with such amount of data, a new emerging technology 'Big Data Analytics' is explored and employed for analyzing consumer behaviors and searching their information needs. Specifically, this paper proposes a Big Data application framework for analyzing consumer behaviors by using topological data structure, co-occurrence methodology and Markov chain theory. First, the consumer related data is translated into a topological data structure. Second, using topological relationships, a co-occurrence matrix is formed to deduce Markov chain model for consumer behavior analysis. Finally, some simulation results are shown to confirm the effectiveness of the proposed framework. © 2013 IEEE.
-
Archery sight-system by magnetic sensors for visually impaired persons
Sasayama T., Oka I., Ata S., Zin T., Watanabe H., Sasano H.
International Conference on Advanced Technologies for Communications 559 - 562 2013.12
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：International Conference on Advanced Technologies for Communications
The archery sight-system is developed for visually impaired persons. The system is based on the magnetic directional sensors for earth magnetism. Both azimuth and elevation angles to the target are measured by the sensor equipped on the archery bow. The sensed information is transmitted to the receiver, which is located near the archer. The sound interface in the receiver transforms the sensed information to the hearing information, and the archer gets this hearing information by the wireless headphones. The sensor system is light-weight of 34g, and operates for about 2 hours by two SR44 button batteries. Shooting experiments by archers with eye-mask show that the system is useful for the visually impaired persons to play archery. © 2013 IEEE.
-
A stochastic model for measuring popularity and reliability in social network systems
Zin T., Tin P., Toriu T., Hama H.
Proceedings - 2013 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2013 462 - 467 2013.12
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Proceedings - 2013 IEEE International Conference on Systems, Man, and Cybernetics, SMC 2013
Popularity and reliability information are crucial ingredients of today social networking systems such as Facebook, LinkedIn, YouTube, Twitter, and so on. In this paper, we propose a stochastic model for measuring and ranking popularity and reliability information in social networks. Specifically, by using the relationships between co-occurring users, we model a Markov chain for reliability measures and a queuing model for popularity measures based on time, space and scenarios. We then form a convex combination or fusion of the two measures to compute the Integrated Global Rank for social networks systems. Finally, we present some illustrative simulation results by using the social networks data collected from Twitter and YouTube. Experimental study indicates the effectiveness of proposed ranking algorithm in terms of better search results. © 2013 IEEE.
DOI： 10.1109/SMC.2013.84
-
Similar image retrieval system reflecting user intention using shape descriptors
Iwanaga T., Zin T., Hama H., Toriu T., Tin P.
ICIC Express Letters 7 ( 4 ) 1425 - 1430 2013.2
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters
Most existing image search systems cannot always provide satisfactory information to the users because relevant images differ according to user or requirement. It is a challenging task to establish an image search system which effectively and properly takes users demands into account. Thus in this paper we propose a new technique for searching images in large databases to meet the user's demands. The proposed system reflects user demand by user selection from choices showed by system. In our system, we used two shape descriptors: the region-based and contour-based. Moreover, we confirmed the effectiveness of the proposed system by comparing with the ordinary system. Experimental results show that our approach is very powerful and this also leads to satisfy the user satisfaction. © 2013 ICIC International.
-
A two dimensional correlated random walk model for visual tracking
Zin T., Tin P., Toriu T., Hama H.
ICIC Express Letters 7 ( 5 ) 1501 - 1506 2013.2
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters
This paper proposes a novel approach to visual tracking by using a sequential two dimensional correlated random walk model. In this approach we use Markov chain sampling techniques to estimate the most likely parameters including the spatiotemporal characteristics of object motions in the setting of occlusions and clutters. Based on the estimated parameters, a discriminant probability matrix is established to extract the target foreground object. The probability matrix is updated by an incremental maximal principle to handle the appearance variations of the target and background. The fundamental equation which describes a two dimensional random walk with non-uniform jump is derived by using the updated probability matrix. Then, it is applied to a visual analysis of human motion tracking problems. The efficiency of the proposed method is confirmed by our experimental results obtained in complex scenario. © 2013 ISSN 1881-803X.
-
A stochastic petri net framework for human behavior analysis in surveillance video Reviewed
Pyke Tin, Thi Thi Zin, T. Toriu, H. Hama
ICIC Express Letters 7 ( 5 ) 1675 - 1680 2013.2
Language：English Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters
Analyzing human behaviors and detecting unusual events is an essential task in surveillance applications for prevention of security threats. In order to carry out this task, a proper way of representing an event and an efficient method of recognizing such events are required. In this paper, we propose a stochastic Petri Net framework for both representation and recognition of unusual events involving suspicious people and suspicious objects. We first use a formal low level image processing method for background subtraction and then automatically map the extracted foregrounds into stochastic Petri Net framework for event representations. In the next we generate reachability graphs to describe human behavior models concerning with specific event nets. Recognition process is then carried out by moving tokens through the Petri Nets. In particular we investigate some unusual human-object-human interactions such as theft detection, exchange of objects and two persons fighting. The experimental results show the effectiveness of our approach to human behavior analysis problems attaining high accuracy rates. © 2013 ISSN 1881-803X.
-
A series of action recognition using HMM and probability representation based on binned features set
Sugimoto M., Zin T., Toriu T., Tin P.
ICIC Express Letters 7 ( 4 ) 1419 - 1424 2013.2
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters
We present a more practical approach for human action recognition in video image sequences. The methodology is the combination of Hidden Markov Model (HMM) and probability representation based on binned features set, including silhouette and feature extraction. This enables to trace a person's action consecutively. For this aim, we define seven types of fundamental actions such as walking and running. The proposed system consists of two major steps: the training of HMM parameters and the testing video sequences for a series of actions. The experimental results show that the system can recognize the seven actions with high accuracy rate. Simultaneously, they also show HMM produces good performance to serial human action recognition for given the features set and training sequences. © 2013 ICIC International.
-
Image processing approach to automatic scoring system for archery targets
Zin T., Oka I., Sasayama T., Ata S., Watanabe H., Sasano H.
Proceedings - 2013 9th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2013 259 - 262 2013
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Proceedings - 2013 9th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2013
This paper proposes an image processing approach to automatic scoring system for archery targets. Specifically, the proposed system first performs Morphological operations on the target image to obtain thick boundaries of the arrow hits and then the image segmentation process is applied to segment the target area by using color and shape features. In doing so, a dynamic thresholding is applied to handle the possible effects of illumination variations and the bull's eyes or the target concentric circles are segmented by using geometrical properties of circles and ellipses to calculate hit scores. Thus the system can have a capability of scoring inside and outside the concentric circles separately. The proposed method is tested by using our own video sequences taken in the self-conducted experiments performed by four archers on two different days shooting 6 arrows. The test results show that the proposed scoring system is promising with an accuracy rate of 100%. © 2013 IEEE.
-
An integrated framework for disaster event analysis in big data environments
Tin P., Zin T., Toriu T., Hama H.
Proceedings - 2013 9th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2013 255 - 258 2013
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Proceedings - 2013 9th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2013
Today world has witnessed the catastrophic consequences of natural and man-made disasters are demanding the urgent need for more research to advance fundamental knowledge and innovation for disaster prevention, mitigation and management. At the same time, the world is in the age of the Big Data revolution which holds the potential to mitigate the effects of disaster events by enabling access to critical real time information. Thus, in this paper an integrated framework for analyzing disaster events by using the Big Data analytics is proposed. The proposed framework shall address three key components to perform data organization, data integration and analysis, information presentation to users by utilizing Big Data with respect to disaster events. In doing so, the paper shall create a disaster domain-specific search engine using co-occurring theory and Markov chain concepts for preparing impacts of disaster attacks to make the society better aware of the situations. Specifically, stochastic clustering with constraints is used to automatically extract disaster events by defining the set of structural attributes. Some illustrative simulations are shown by using Big Data sources for the Great East Japan earthquake, tsunami and nuclear disaster events of 2011. © 2013 IEEE.
-
A Markov chain model for image ranking system in social networks
Zin T., Tin P., Toriu T., Hama H.
Proceedings of SPIE - The International Society for Optical Engineering 9027 2013
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Proceedings of SPIE - The International Society for Optical Engineering
In today world, different kinds of networks such as social, technological, business and etc. exist. All of the networks are similar in terms of distributions, continuously growing and expanding in large scale. Among them, many social networks such as Facebook, Twitter, Flickr and many others provides a powerful abstraction of the structure and dynamics of diverse kinds of inter personal connection and interaction. Generally, the social network contents are created and consumed by the influences of all different social navigation paths that lead to the contents. Therefore, identifying important and user relevant refined structures such as visual information or communities become major factors in modern decision making world. Moreover, the traditional method of information ranking systems cannot be successful due to their lack of taking into account the properties of navigation paths driven by social connections. In this paper, we propose a novel image ranking system in social networks by using the social data relational graphs from social media platform jointly with visual data to improve the relevance between returned images and user intentions (i.e., social relevance). Specifically, we propose a Markov chain based Social-Visual Ranking algorithm by taking social relevance into account. By using some extensive experiments, we demonstrated the significant and effectiveness of the proposed social-visual ranking method.
DOI： 10.1117/12.2042621
-
Knowledge based social network applications to disaster event analysis
Zin T., Tin P., Hama H., Toriu T.
Lecture Notes in Engineering and Computer Science 2202 279 - 284 2013
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Lecture Notes in Engineering and Computer Science
Today online social networking platforms such as Facebook, Flicker, Twitter, and YouTube often serve a breaking news role for natural disasters. The role of these social networks has significantly increased following the recent disasters around the world; the 2010 Philippine typhoon, the 2010 Haiti earthquake, the 2011 Brazil flood, and the 2011 Japan earthquake and tsunami. Moreover, these platforms are among the first ones to help communicate the news to a large mass of people since they are visited by millions of users regularly. In such emergency situations, detecting and analyzing hot spots or key events from the pool of information in the social networks are of major concerns in assessing the situation and in decision making. In this paper, a knowledge based event analysis framework for automatically analyzing key events is proposed by using various social network sources in case of disasters. In doing so, some mathematical modeling techniques of branching processes and Markov chain theory are explored and employed to investigate how news about these disasters spreads on the social networks and how to extract trust and reliable key information. Specifically the abnormal or suspicious topics and important events within various social network platforms are analyzed by using a set of selected messages and visual data. Finally some illustrative sample results are presented based on a limited datasets of YouTube and Twitter in the case of March 11, 2011 Japan Earthquake and Tsunami.
-
Visual behavior analysis tool for consumer video surveillance
Zin T., Tin P., Toriu T., Hama H.
1st IEEE Global Conference on Consumer Electronics 2012, GCCE 2012 718 - 719 2012.12
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：1st IEEE Global Conference on Consumer Electronics 2012, GCCE 2012
Consumers video surveillance systems are now being used not only for security reasons but also for better understanding consumer behaviors. In this paper, we propose a new visual behavior analysis tool for consumer video surveillance systems. This tool can be embedded in consumer videos to automatically detect and analyze unusual events. The proposed tool is developed by using a special type of Gamma Markov chain for background modeling and Petri Nets for object classification. We present some experimental results to show the effectiveness of the proposed system which will be leading to new visual behavior analysis tools for the consumers. © 2012 IEEE.
-
Conceptual vision keys for consumer product images
Zin T., Tin P., Toriu T., Hama H.
1st IEEE Global Conference on Consumer Electronics 2012, GCCE 2012 435 - 436 2012.12
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：1st IEEE Global Conference on Consumer Electronics 2012, GCCE 2012
In the consumer world, the ever growing image repositories in online shopping, consumer products images, consumer photos and video collections have resulted great demand of a system which can accurately retrieve similar images from image database. For this purpose, we propose a new concept of vision key for retrieving consumer product images. In our system, rather than considering an image as a whole, we consider it as a set of regions or sub-images with completely different semantic meanings. By using the properties of equivalence classes in the Markov chain, we first perform image segmentation and initial pixel grouping process. We then establish vision keys by using a Markov stationary feature. Finally, in the retrieval phase, users can interactively search candidate images which contain vision keys. In order to confirm the efficiency of our proposed method, we present the experimental results achieving on higher accuracy rates. © 2012 IEEE.
-
A series of stochastic models for human behavior analysis
Zin T., Tin P., Toriu T., Hama H.
Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics 3251 - 3256 2012.12
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Conference Proceedings - IEEE International Conference on Systems, Man and Cybernetics
In this paper, we propose a new approach to analyze human behaviors by using a series of stochastic models composed of a Bivariate Gamma Markov model, a two dimensional correlated Random Walk model and a finite state Markov Chain model. Specifically, the proposed method contains three modules namely: (i) image analysis module, (ii) probability analysis module and (iii) event analysis module. We model each module by a special type of stochastic processes forming a series of stochastic models for the complete behavior analysis system. This approach is more effective in utilizing modular stochastic models to describe complex behavior patterns. By assembling these modular models in a series we can design a robust model for the analysis of human behaviors. The feasibility and effectiveness of the proposed method are tested on two different datasets: a self-collected dataset and PETS 2006 dataset. © 2012 IEEE.
-
A novel probabilistic video analysis for stationary object detection in video surveillance systems
Zin T., Tin P., Toriu T., Hama H.
IAENG International Journal of Computer Science 39 ( 3 ) 295 - 306 2012.9
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：IAENG International Journal of Computer Science
In this paper, we propose a novel probabilistic approach for detecting and analyzing stationary objects driven visual events in video surveillance systems. This approach is based on a newly developed background modeling technique and an adaptive statistical sequential analysis method. For background modeling part, we use the concepts of periodic Markov chain theory producing a new background subtraction method in computer vision systems. We then develop an object classification algorithm which can not only classify the objects as stationary or dynamic but also eliminate the unnecessary examination tasks of the entire background regions. Finally, this paper introduces a sequential analysis model based on exponent running average measure to analyze object involved events such as whether it is either abandoned or very still person. In order to confirm our proposed method we present some experimental results tested on our own video sequences taken in international airports and some public areas in a big city. We have found that the results are very promising in terms of robustness and effectiveness.
-
Motion-compensated inter-frame subtraction based on self-organized internal representation Reviewed
T. Toriu, Thi Thi Zin, H. Hama
ICIC Express Letters 6 ( 4 ) 905 - 910 2012.4
Language：English Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters
In a previous paper, we introduced a time evolution operator and based on it proposed an unsupervised learning algorithm for self-organizing an internal representation of ego-motion. In this paper, we propose a method to predict the image at the next instance using the time evolution operator generated on the basis of the internal representation of ego-motion. In addition, we propose a method of motion-compensated inter-frame subtraction. By subtracting the predicted image at the next instance from the true image, we can obtain an image that has high intensity in the region of the moving object. This method is effective even if the camera itself is in motion. We show the results of the experiments conducted using a randomly synthesized image and the real image. ICIC International © 2012.