Papers - THI THI ZIN
-
Visual analysis framework for two-person interaction
Zin T., Kurohane J.
2015 IEEE 4th Global Conference on Consumer Electronics, GCCE 2015 519 - 520 2016.2
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：2015 IEEE 4th Global Conference on Consumer Electronics, GCCE 2015
© 2015 IEEE. In this paper, a novel approach to two person interaction method is presented in which pose representation is based on the feature of silhouette images. Today human activity recognition and analysis has a tremendous potential to impact a wide range of applications from surveillance to human computer interfaces to content based video retrieval. Specifically, the proposed method makes use of human silhouettes to classify actions and interactions of human present in a scene video. The classes of interactions will include punching, pushing, kicking, hand-shaking, and hugging. Moreover, the detected interactions are further divided into violence or non-violence so that a suitable security measures would be taken. To confirm the validity of the proposed method, the experimental results are carried out by using the publicly available dataset.
-
Block based approach for key frame extraction on large video sequences
Tin M., Myo N., Khin M., Zin T.
ICIC Express Letters, Part B: Applications 7 ( 12 ) 2713 - 2717 2016
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters, Part B: Applications
© 2016 ISSN 2185-2766. Video segmentation and key frame extraction are very important in real world video systems. Key frames are essential to analyze on large amount of video frame sequences. This paper emphasizes surveillance video and the aim is to extract some meaningful key frames from long video sequences. This purpose is to reduce weak transaction frames in a large video stream by using block base algorithm and cluster the transition video shots. Kullback-Leible divergence method is used in key frame extraction for strong transition video shot. For weak transition video shot, the system will find three candidate key frames and they are compared. Key frames are meaningful frames for video sequences. These frames which represent video streams can be analyzed. Duplicated key frames from the video stream are analyzed in order to be extracted from different shots. Finally, key frames have many assets such as stability, accuracy, and summarize information for a large video.
-
A novel research topic ranking system in academic networks
Zin T., Tin P., Hama H.
Advances in Intelligent Systems and Computing 387 361 - 368 2016
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Advances in Intelligent Systems and Computing
© Springer International Publishing Switzerland 2016. In today world, various types of communication networks such as academic, social, technological, business and etc. come into front. All of the networks are continuously growing and expanding in volume, velocity and variety like popular platform Big Data. Among them, the academic networks such as Alumni, Research Gate, Student Network, Teacher Network and so on provide a powerful abstraction of the academic structure and dynamics of diverse kinds of inter personal academic activities and interaction. Generally, the academic network contents such as research findings and educational concepts are created and consumed by the influences of all different academic navigation paths that lead to the challenging research issues. Therefore, identifying important and researcher relevant refined structures such as new research topics information or academic communities become major factors in modern decision making world.. In this paper, we propose a novel research topic ranking system in academic networks by using the research data relational graphs from academic media platform jointly with educational data to improve the relevance between research topics and researchers intentions (i.e., academic relevance). Specifically, we propose a stochastic model based Academic-Research Topic Ranking algorithm by taking academic value into account. By using some extensive experiments, we demonstrated the significant and effectiveness of the proposed academic-research topic ranking method.
-
Zin T., Lin J., Pan J., Tin P., Yokota M.
Advances in Intelligent Systems and Computing 387 2016
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Advances in Intelligent Systems and Computing
-
A novel method for product brand ranking in consumer networks Reviewed
Thi Thi Zin, Pyke Tin, H. Hama
Proceedings of The 4th IEEE Global Conf. on Consumer Electronics (GCCE 2015) 328 - 329 2015.10
Language：English Publishing type：Research paper (international conference proceedings)
-
Visual analysis framework for two-person interaction Reviewed
Thi Thi Zin, J. Kurohane
Proceedings of The 4th IEEE Global Conf. on Consumer Electronics (GCCE 2015) 519 - 520 2015.10
Language：English Publishing type：Research paper (international conference proceedings)
-
A color constancy model for non-uniform illumination based on correlation matrix Reviewed
T. Toriu, M. Hironaga, H. Kamada, Thi Thi Zin
Proceedings of the Tenth International Multi-Conference on Computing in the Global Information Technology 40 - 46 2015.10
Language：English Publishing type：Research paper (international conference proceedings)
-
Cow identification by using shape information of pointed pattern
K. Sumi, I. Kobayashi, Thi Thi Zin
Genetic and Evolutionary Computing (Proceedings of the 9th Intl. Conf. on Genetic and Evolutionary Computing) 2 273 - 280 2015.8
Language：English Publishing type：Research paper (scientific journal)
-
A novel research topic ranking system in academic networks Reviewed
Thi Thi Zin, Pyke Tin, H. Hama
Genetic and Evolutionary Computing (Proceedings of the 9th Intl. Conf. on Genetic and Evolutionary Computing) 1 361 - 368 2015.8
Language：English Publishing type：Research paper (international conference proceedings)
-
A Matrix-Geometric Method for Web Page Ranking Systems Reviewed
Thi Thi Zin, Pyke Tin, H. Hama, T. Toriu
Journal of Information Hiding and Multimedia Signal Processing 6 ( 4 ) 639 - 647 2015.7
Language：English Publishing type：Research paper (scientific journal)
-
A Novel Hybrid Approach to Image Ranking System Reviewed
Pyke Tin, Thi Thi Zin, T. Toriu and H. Hama
ICIC International, ICIC Express Letters (Part B: Applications) 6 ( 3 ) 743 - 748 2015.3
Language：English Publishing type：Research paper (scientific journal)
-
A Triplet Markov Chain Model for Loitering Behavior Detection Reviewed
Thi Thi Zin, Pyke Tin, H. Hama, T. Toriu
ICIC International, ICIC Express Letters (Part B: Applications) 6 ( 3 ) 613 - 618 2015.3
Language：English Publishing type：Research paper (scientific journal)
-
Image smoothing using a metric tensor for an affine invariant scale space
Toriu T., Zin T., Hama H.
2014 4th International Conference on Image Processing Theory, Tools and Applications, IPTA 2014 2015.1
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：2014 4th International Conference on Image Processing Theory, Tools and Applications, IPTA 2014
© 2014 IEEE. This paper proposes a new image smoothing method using a metric tensor for affine invariant scale space. In the field of image processing and recognition, Gaussian filtering is a common procedure for image smoothing. For example, scale space construction based on Gaussian filtering is sometimes used as a preprocessing of various image processing tasks. However Gaussian filtering is not affine invariant. This paper proposes a new method for image smoothing that is invariant under such affine transformation that does not change the area of any region in the image. It is shown that a scale space representation can be constructed collaterally with the image smoothing. Experimental results show that the proposed method is almost never affected by affine transformation different from usual Gaussian filtering. In the proposed method, processing results are expected to be not affected much by variation of the viewpoint.
-
Cow identification by using shape information of pointed pattern Reviewed
K. Sumi, I. Kobayashi, Thi Thi Zin
Advances in Intelligent Systems and Computing 388 273 - 280 2015
Language：English Publishing type：Research paper (scientific journal) Publisher：Advances in Intelligent Systems and Computing
Monitoring cow behavior and performance plays an important role in dairy health and welfare management systems. Due to the increased number of elderly farm workers, the demands for automatic cow monitoring system become a key role. Moreover, the image based technology for a monitoring system is a promising technique because it is relatively low cost and easy to install. In this aspect, the fundamental and important work to be done is to make an identification of individual cows with high accuracy. Thus in this paper, a simple and effective method for cow identification is introduced by using a modified background subtraction method and histogram based decision process. Specifically, the painted-marks are placed on all black-haired cows and video images are taken. Then the marked region is extracted by using the proposed background subtraction method and histogram based features. Finally, the identification process is performed and some experimental results are shown by using self-collected database taken in the University dairy farm.
-
A triplet Markov chain model for loitering behavior detection
Zin T., Tin P., Hama H., Toriu T.
ICIC Express Letters, Part B: Applications 6 ( 3 ) 613 - 618 2015
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters, Part B: Applications
© 2015 ICIC International. Currently, a sizable number of monitoring systems have been established in public areas where the security factors are sensitive to discourage and prevent crimes. One of applications in such places is to detect loitering people since they can lead to various forms of criminal actions such as drug dealing, pick-pocking, theft and many others. To address this problem, we propose a new concept of Triplet Markov Chain model for detecting loitering people in public areas. After having extracted the foreground object, people are recognized amongst the foreground objects. Then they start being embedded into a Markov Chain by using three types of processes, namely, observing process, state process and underlying process. Finally, if the targeted state remains in the scene longer than the user-defined loitering time threshold then the loitering behavior is detected. The proposed method has been confirmed by using our own video sequences and PETS 2007 datasets.
-
A novel hybrid approach to image ranking system Reviewed
Pyke Tin, Thi Thi Zin, T. Toriu, H. Hama
ICIC Express Letters, Part B: Applications 6 ( 3 ) 743 - 748 2015
Language：English Publishing type：Research paper (scientific journal) Publisher：ICIC Express Letters, Part B: Applications
Today image ranking systems mostly aim at bridging visual images and user intentions through content-based image retrieval concepts. However, it is still far from satisfactory to narrow down the gap between the outcome images and users expectations. Many researchers recognized that by using visual content or textual content alone it is not easy to identify the user’s interest. In this paper, we propose a novel hybrid approach to image re-ranking system which integrates both visual information of an image and user concepts by forming image interest groups. Specifically, we first establish two stochastic matrices namely, image-content matrix and user-intent matrix, and then the two matrices are combined as a convex combination to form a new user-intent-visual matrix. Finally, the stationary distribution of the matrix is used as an image ranking system so that the users can obtain the desired images from the image datasets easily and effectively. Some experimental results are given to confirm our proposed method.
-
A new look into Web page ranking systems
Zin T., Tin P., Hama H., Toriu T.
Advances in Intelligent Systems and Computing 329 343 - 351 2015
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Advances in Intelligent Systems and Computing
© Springer International Publishing Switzerland 2015. This paper proposes a new way of looking into Web page ranking systems by using some concepts of queuing theory in operations research and stochastic water storage theory in hydrology. Since both theories queuing and stochastic water storage are rich in technology as well as application aspects, the new look in this paper may lead to new directions in Web page ranking systems and related research areas. In doing so, first this paper draws some analogies between a Web page ranking system and theory of queues. Then it shows how a Web page ranking system can be tackled to reduce current obstacles by using queuing theory techniques. In the second, a Web page ranking system is modeled as a framework of stochastic water storage theory to derive a list of Web page rankings. Third and finally, the outcome results of rankings obtained by using the proposed two theories queuing theory and stochastic water storage are compared and analyzed analytically as well as experimentally. The experimental results show the proposed new look is promising for establishing a new research area which can improve the current situations and difficulties occurred in search engines and their ranking systems in particular and some problems in World Wide Web as a whole.
-
A matrix-geometric method for web page ranking systems
Zin T., Tin P., Hama H., Toriu T.
Journal of Information Hiding and Multimedia Signal Processing 6 ( 4 ) 639 - 647 2015
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Journal of Information Hiding and Multimedia Signal Processing
© 2015 ISSN. An algorithmic approach which is known as Matrix-Geometric method is an efficient and popular in solving various types of stochastic models especially for queueing and water storage models. This paper proposes a modified Matrix-Geometric method for calculating random web page ranks in World Wide Web (WWW) by considering web page ranking system as a stochastic model in random environments. Since the Matrix- Geometric approach is very powerful in variety of stochastic models, it may develop new promising directions in Web Page Ranking Systems and related research areas. In order to do so as first step, a web page ranking system is modeled in the frameworks of queueing models. Then an attempt is made to reduce the obstacles occurred in the problems of web page ranking systems. Similarly, in the second step, a Web page ranking system is modeled as a framework of stochastic water storage theory to derive a list of Web page rankings by using Matrix-Geometric method. Some comparison results are presented to confirm the efficiency of the proposed methods. The experimental results shows the pro- posed approach is promising for establishing a new research area which can improve the current situations and difficulties occurred in search engines and their ranking systems in particular and some problems in WWW as a whole.
-
Zin T., Lin J., Pan J., Tin P., Yokota M.
Advances in Intelligent Systems and Computing 388 2015
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Advances in Intelligent Systems and Computing
-
A cluster based ranking framework for multi-typed information networks
Tin P., Toriu T., Zin T., Hama H.
Proceedings - 2014 10th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2014 415 - 418 2014.12
Language：Japanese Publishing type：Research paper (scientific journal) Publisher：Proceedings - 2014 10th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, IIH-MSP 2014
© 2014 IEEE. A multi-typed information network is an information network which contains multiple types of objects having actions and interactions between each other. Although many studies on single typed information network haven been found in the literature, only a little has been known concerning with multi-typed information networks. On the other hand, multiple type information networks are ubiquitous and forming an important component of modern information infrastructure. Thus, in this paper we propose a new method to give a better understanding of information networks and their properties. Specifically we propose a new cluster based ranking system for multi-typed information networks. In this aspect, ranking evaluates objects of information networks based on some mathematical ranking function which illustrates the characteristic of objects with which any two objects of the same type can be compared by qualitatively. Moreover, clustering group objects is based on a certain measure such that similar objects are in the same cluster whereas dissimilar objects are in different clusters. Then the ranking and clustering processes are integrated to extract insight overall views of information networks, so that the integrated method can be widely applied in different information network settings. Our experiments using DBLP datasets can generate good informative clusters producing reliable ranking system.