Yu Tzu Lin

Real-time eye-gaze estimation using a low-resolution webcam

Eye detection and gaze estimation play an important role in many applications, e.g., the eye-controlled mouse in the assisting system for disabled or elderly persons, eye fixation and saccade in psychological analysis, or iris recognition in the security system. Traditional research usually achieves eye tracking by employing intrusive infrared-based techniques or expensive eye trackers. Nowadays, there are more and more needs to analyze user behaviors from tracking eye attention in general applications, in which users usually use a consumer-grade computer or even laptop with an inexpensive webcam. To satisfy the requirements of rapid developments of such applications and reduce the cost, it is no more practical to apply intrusive techniques or use expensive/specific equipment. In this paper, we propose a real-time eye-gaze estimation system by using a general low-resolution webcam, which can estimate eye-gaze accurately without expensive or specific equipment, and also without an intrusive detection process. An illuminance filtering approach is designed to remove the influence from light changes so that the eyes can be detected correctly from the low-resolution webcam video frames. A hybrid model combining the position criterion and an angle-based eye detection strategy are also derived to locate the eyes accurately and efficiently. In the eye-gaze estimation stage, we employ the Fourier Descriptor to describe the appearance-based features of eyes compactly. The determination of eye-gaze position is then carried out by the Support Vector Machine. The proposed algorithms have high performances with low computational complexity. The experiment results also show the feasibility of the proposed methodology.

Information behavior and value creation potential of information capital: Mediating role of organizational learning

This work draws on recent literature concerning information capital and organizational learning to explore how this may affect information behavior and value (IBV) of the healthcare industry in an emerging economy. Whether organizational learning mediates the relationships between information capital and both information behavior and information value (IBV) is also examined. Additionally, a parsimonious and plausible empirical model is proposed to examine the links among contextual variables. Data are collected from full-time medical staff of medical centers in Taiwan, including physicians and nurses. To clarify the relationships among these variables, structural equation modeling (SEM) is conducted to examine the model fit and three hypotheses. The SEM results clearly demonstrate the mediating role of organizational learning in the impact of information capital on both information behavior and information value. Consequently, organizational learning leads to appropriate information behavior and information value (that is, information sharing and information integrity, respectively) only when integrated with organizational learning and information capital for healthcare organizations.

Indexing and Teaching Focus Mining of Lecture Videos

This paper proposes an indexing and teaching focus mining system for lecture videos recorded in an unconstraint environment. The slide structure can be reconstructed by an edge-based shot change detection algorithm. Besides, the teaching focus can be extracted according to instructor’s behavior, including the gesture, the lecture time for each slide, and the speech speed. Experiment results show the feasibility of the proposed method, that is, the slide shots can be correctly detected even if the illumination conditions is variant or the slides are obstructed by the instructor or students, and the teaching focus can be well extracted to provide learners an efficient way to study.

Exploring the effects of employing Google Docs in collaborative concept mapping on achievement, concept representation, and attitudes

This study investigated the effectiveness of using Google Docs in collaborative concept mapping (CCM) by comparing it with a paper-and-pencil approach. A quasi-experimental study was conducted in a physics course. The control group drew concept maps using the paper-and-pencil method and face-to-face discussion, whereas the experimental group employed Google Docs to create concept maps and used Google Chat for discussion. Learning achievements, physics concept representation, attitudes toward science, and attitudes toward CCM were examined to identify the differences between the two groups and the effectiveness of Google Docs. The results suggested that the use of Google Docs did not significantly affect physics achievement. However, Google Docs fostered physics concept representation and enhanced attitudes toward science. Moreover, the Google Docs group tended to learn more collaboratively. The students in this group also agreed with the superiority of observation and modification functions provided by the real-time co-editing mechanism and revision history.

Structuring and analyzing low-quality lecture videos

This paper presents a lecture video structuring and analysis scheme to provide students an efficient way to access the lecture content. Instead of using color-based or histogram-based methodologies, we propose a new edge-based shot boundary detection algorithm to accurately rebuild the slide structure. The proposed approach can successfully resist the unwanted influences induced from the variant illumination condition and occlusions. Besides, original slide content can be extracted excluding any obstruction by using human removing techniques. Furthermore, the teaching focus is analyzed so that this system becomes more useful for learning.

Practical fingerprinting system for images

This paper presents a feasible system for image fingerprinting. Two important problems of fingerprinting are the design collusion-secure codes and the watermarking strategies for the fingerprints. We address these problems by proposing hierarchical fingerprint codes and a content adaptive watermarking scheme. It is challenging to find a practical collusion-secure code for fingerprinting because of the trade-offs between the fingerprint codes length, its collusion resiliency, and the size of the customer base. Therefore, instead of pursuing a high-performance code, we construct a new fingerprint code by concatenating existing codes using a hierarchical structure. The resulting code is efficient in decoding, much shorter than existing codes for similar parameters, and thus practical for real systems. In the fingerprinting, the codes are often very long if the applicable colluder-resisting ability and large customer bases are needed. The resulting acute degradation of fingerprinted content may be unacceptable. Hence the long-length watermark must be appropriately embedded into images to preserve the image quality. We design a strength-decision algorithm, on the basis of neural networks, to adaptively embed the long-length fingerprints with suitable magnitudes into different image regions. This adaptive watermarking technique maintains the combination of robustness and imperceptibility without the effort of dealing with visual models. In addition, we analyze the disturbance due to collusion attacks on images and propose an optimization algorithm that can select better embedding positions to resist collusion attacks and preserve acceptable transparency of the watermark according to different multimedia contents. Experimental results show the high detection correctness of the traitor tracing. It implies that our fingerprinting system, constructed by applying a hierarchical fingerprint code to the content-adaptive watermarking scheme, is effective for multimedia application.

Removable Visible Watermarking in JPEG Compression Domain

The purpose of this research lies in the development of a user-key dependent removable visible watermarking scheme in JPEG compression domain. The protected images are embedded with the visible watermarks, and further compressed by JPEG compression. The visible watermarks only can be successfully removed by authorized user with the private key. Experimental results show the proposed scheme can recover images to the extent of JPEG-compression qualities.

Learning-Focused Structuring for Blackboard Lecture Videos

Since blackboards are the standard in the classrooms and are still used today, blackboard lecture videos are common in the lecture video recordings. However, it has been known that content-based blackboard lecture video analysis is challenging and thereby rarely be touched upon in the field of semantic computing. In this paper, we proposed a new structuring method for blackboard lecture videos by estimating the learning focus that learners should pay more attention to. Both visual and aural analysis for blackboard lecture videos are utilized and integrated to develop a learning-focused attention model. As for the visual analysis, the fluctuation of written lecture content on the blackboard and the posture of lecturers are analyzed. On the other hand, the speech of lecturers is used for aural analysis. Finally, a learning-focused attention curve can be generated by fusing multiple attention models. In a sense, the values of the learning-focused attention reflect the strength of attention or semantics that the learners should pay to the blackboard lecture video and can be used for indicating the importance of the extracted lecture content at the corresponding time. Therefore, learners can easily access the blackboard lecture video with good flexibility to find what the lecture content they should understand and video frames to watch from the well-structured video. The experimental results show that the proposed method can effectively structure blackboard lecture videos and extract the lecture content with associated learning-focused attention values.

Traceable multimedia fingerprinting based on the multilevel user grouping

This paper presents a traceable fingerprinting scheme based on a multilevel user grouping strategy. In the proposed code structure, coding efforts for groups of users are apportioned to different levels of the code. Instead of making efforts to find one ldquohigh-performancerdquo traceability code, we try to derive a generalized framework of fingerprints construction which hierarchically constructs a traceability code having ldquohigh-performancesrdquo by composing several ldquomoderate-performancerdquo traceability codes together. Both theoretical analyses and experimental results are provided to show good performances of the proposed fingerprinting system, that is, it can serve a large amount of users with a much shorter code length and high traceability.

Computational Thinking Curriculum for K-12 Education -- A Delphi Survey

The purpose of this study is to design a computational thinking curriculum standard for K-12 education. The Delphi technique was employed to collect different views and derive consensus from a panel of thirteen experts, including computer scientists, computer science educators, K-12 computer teachers, and industry experts. The first draft of Delphi survey questionnaire, consisting of nine themes (problem solving, problem decomposition, algorithms, data representation, data analysis, modeling and simulation, abstraction, automation, and others) and 60 competence indicators, was developed based on our investigation of the nature of computational thinking. After three rounds of survey and a final round-table discussion, the expert panel derived 49 essential competence indicators (13 for grades K to 6, nine for grades 7 to 9 and 27 of grades 10 to 12) and eight optional competence indicators (one for grades 7 to 9 and seven for grades 10 to 12). According to the results, the core ability and training of computational thinking are different in grades. In grades K to 6, students develop logical thinking and problem solving skills in personal and family life. In grades 7 to 9, students must understand the basic concepts of programming and write basic programs. Finally, in grades 10 to 12, they will use high-level skills to apply the concepts of abstraction, modeling, and structured problem decomposition to solve problems in different fields. The results serve as useful references for developing Taiwans new K-12 computing curriculum.