Skip to main content

Über dieses Buch

This book constitutes the proceedings of the 13th International Symposium on Smart Graphics, SG 2015, held in Chengdu, China, in August 2015.

The 17 full and 3 short papers presented in this volume were carefully reviewed and selected from 35 submissions. They are organized in topical sections named: graphics, sketching and visualization, automation and evaluation, image processing, and posters and demo session.





Designing Mini Block Artwork from Colored Mesh

Mini block artwork is a kind of well abstracted low-resolution block construction with aesthetically pleasing block layout. Similar with previous LEGO constructions, mini block artwork requires strong interconnection among blocks, i.e., a stable layout. However, what make mini block artwork different are the new requirements on highly abstracted shapes and colors and the regularity in block layout considering symmetry in the model itself. We focus on these requirements by first integrating quantization of colors into abstraction. We further explore layout generation method satisfying both stability and symmetry to support our prototype design system. Mini block artwork generated using different methods are evaluated on both stability and symmetry of the block layout. To facilitate a justified and discriminating layout comparison using stability, though we consider factors similar to classical heuristics, we experimentally optimize the weight of each factor for mini block artwork.
Man Zhang, Yuki Igarashi, Yoshihiro Kanamori, Jun Mitani

Interactive and Procedural Modeling of Featured Chinese Architectures

The traditional Chinese garden contains many types of tings, corridors, walls, etc. For artists, it is tedious work to model these kinds of featured Chinese architecture due to the strict and complex construction rules. We propose an interactive and procedural tool to modeling featured Chinese architectures that appear in the Chinese garden. Based on the previous research about modeling basic structures of Chinese architecture, we extend to model more featured Chinese architectures, such as double-eave ting, combined ting, corridor, and wall effectively and efficiently, and combine them into a complete Chinese garden. By adjusting the overlapped components, we can combine two single tings into a combined ting. By modifying ting’s structure, we can construct a variant of corridors or garden walls upon few input parameters. In addition, the result 3D model can be exported in different LODs, making the use of the model more practicable and flexible. As experimental results shown, complex 3D models of a Chinese garden with several different featured Chinese architectures can be created in minutes.
Chun-Yen Huang, Yang-Siu Sheng, Wen-Kai Tai

Screen Space Hair Self Shadowing by Translucent Hybrid Ambient Occlusion

Screen space ambient occlusion is a very efficient means to capture the shadows caused by adjacent objects. However it is incapable of expressing transparency of objects. We introduce an approach which behaves like the combination of ambient occlusion and translucency. This method is an extension of the traditional screen space ambient occlusion algorithm with extra density field input. It can be applied on rendering mesh objects, and moreover it is very suitable for rendering complex hair models. We use the new algorithm to approximate light attenuation though semi-transparent hairs at real-time. Our method is implemented on common GPU, and independent from pre-computation. When it is used in environment lighting, the hair shading is visually similar to however one order of magnitude faster than existing algorithm.
Zhuopeng Zhang, Shigeo Morishima

Optimizing Aesthetic-Based Photo Retargeting

Photography is an art based on light. Modern cameras have auto exposure and auto focus functions, so that we can easily take photos with right exposure and focus setting. However, a nice photo depends not only on light used, its composition is also an important factor. We therefore exploit the state-of-the-art retargeting technique to automatically adjust photos for conforming the aesthetic composition. Our approach can use suitable retargeting techniques, and coordinate the composition of original photos to make photos conform the aesthetic rules. The photo types in our system particularly apply to group photo. To the best of our knowledge, this part has not been explored in existing literature. We analyze the common rules of composition, and propose the rules applied to suit group photos. Besides, the photos are adjusted based on the human face. We believe that using the development of our research, everyone can take an ideal photo.
Damon Shing-Min Liu, Chi-Cheng Huang

Sketching and Visualization


InvisiShapes: A Recognition System for Sketched 3D Primitives in Continuous Interaction Spaces

Continued improvements and rising ubiquity in touchscreen and motion-sensing technologies enable users to leverage mid-air input modalities for intelligent surface sketching into the third dimension. However, existing approaches largely either focus on constrained 3D gesture sets, require specialized hardware setups, or do not deviate beyond surface sketching assumptions. We present InvisiShapes, a recognition system for users to sketch 3D geometric primitives in continuous interaction spaces that explore surfaces and mid-air environments. Our system leverages a collection of sketch and gesture recognition techniques and heuristics and takes advantage of easily accessible computing hardware for users to incorporate depth to their sketches. From our interaction study and user evaluations, we observed that our system successfully accomplishes strong recognition and intuitive interaction capabilities on collected sketch+motion data and interactive sketching scenarios, respectively.
Paul Taele, Tracy Hammond

FloodSight: A Visual-Aided Floodlight Controller Extension for SDN Networks

The Software Defined Networking (SDN) is considered as a primary evolutionary direction for the next generation networks. The OpenFlow protocol enabling SDN decouples the control plane from the data plane, thus complex controlling and management functions are able to be eliminated from switches, resulting in dumb switches with fewer layers and higher forwarding efficiency. Sophisticated controlling and management functions seen in traditional switches are moved to dedicated controllers in an SDN environment. Therefore, controllers play an important role in SDN. Floodlight is a widely used and most accepted controller, which offers a variety of useful functions. However, Floodlight lacks several key features needed in a simulation-oriented SDN environment, especially the GUI-aided configurations. To tackle there problems, we propose FloodSight (Floodlight with Sight) in this paper to assist network administrators and researchers to efficiently prototype and test an SDN network with visual support. FloodSight is a visual-aided Floodlight extension based on Floodlight’s REST APIs. The current version of FloodSight consists of 3 major components: the QoS-aware Topology Viewer, the Topology Maker and the Flow Manager, which offer visual support for QoS-aware topology viewing, topology making and flow manipulation not seen in the original Floodlight controller. The experiments on the FloodSight show that it offers desirable features and feasible performance.
Xi Chen, Dongqi Guo, Wufangjie Ma, Longhui He

From 2D to 3D: A Case Study of NPR and Stereoscopic Cinema

Our interdisciplinary research is dedicated to exploring the boundaries of stereoscopic filmmaking from an unusual viewpoint: we aim at creating 3D non-photorealistic cinema which allows conciliating a stereoscopic pre-visualization that is oriented to ensure visual comfort with concept tests of NPR applied to a 3D film. In this paper we describe the role of pre-visualization in stereoscopic cinema and our preliminary observations and experience of combining 3D cinema with non-photorealistic rendering approaches, from the filmmakers’ point of view.
Victor Fajnzylber, Milán Magdics, Macarena Castillo, Constanza Ortega, Mateu Sbert

Research on Collaborative Visualization Application of Dynamic Monitoring Figure Spot

Dynamic Monitoring is one of the most basic and the most important parts in geographical conditions monitoring, which provides basic data for geographical conditions monitoring and assists local governments with land survey and database updating quickly, accurately and completely. Based on the uniform principle, it can extract various types of figure spot for land using, by the HCI-based methods with the prototype system, through the registration and contrastive analysis of the remote sensing images, the land survey database of last annual, temporary polygons, etc. Binding the characteristics of Big Data environments, the complexity, isomerism and distributives of spatial information system as well as the diversity and personalization of user needs, determine Dynamic Monitoring should have the characteristics of universality and synergy. It also can improve the collaborative visualization efficiency of data service, by establishing a distributed collaborative system and a universal computing environment of spatial information.
Ken Chen, Ping Liao, Fang Wang, Yuchuan Wang, Pengfei Xiao

Automation and Evaluation


Logic Control for Story Graphs in 3D Game Narratives

With the rising popularity of engaging storytelling experiences in gaming arises the challenge of designing logic control mechanisms that can adapt to increasingly interactive, immersive, and dynamic 3D gaming environments. Currently, branching story structures are a popular choice for game narratives, but can be rigid, and authoring mistakes may result in dead ends at runtime. This calls for automated tools and algorithms for logic control over flexible story graph structures that can check and maintain authoring logic at a reduced cost while managing user interactions at runtime.
In this work we introduce a graph traversal method for logic control over branching story structures which allow embedded plot lines. The mechanisms are designed to assist the author in specifying global authorial goals, evaluating the sequence of events, and automatically managing story logic during runtime. Furthermore, we show how our method can be easily linked to 3D interactive game environments through a simple example involving a detective story with a flashback.
Hui-Yin Wu, Tsai-Yen Li, Marc Christie

Usability Evaluation Methods of User Interface Based on Mobile Games Using Fuzzy Methods

The usability evaluation methods of user interface based on mobile games was introduced in the paper. User usability evaluation model about mobile games were established based on user interview and questionnaire. Then, each evaluation item was analyzed and their weights were determined using the Delphi methods and mobile game interface usability index weight level distribution map was obtained. In addition, the methods of gray correlation analysis were proposed to optimize evaluation system model. Meanwhile, the methods of fuzzy comprehensive evaluation were given and the influence factors were found using the principle of maximum degree of membership. In the end, the example provided demonstrates the most satisfied user interface of mobile games is selected and it was proved that the methods proposed is efficient.
Mengtian Cui, Libo Zhu

A Robust Digital Image Watermarking Algorithm Based on DCT Domain for Copyright Protection

With the rapid development of various digital multimedia technologies, the works, such as copying, downloading and publishing digital multimedia works via the Internet, are made very convenient. However, it also brings on a huge challenge to the copyright protection of digital multimedia works. How to effectively protect and authenticate digital multimedia works on the Internet has been a problem to be solved urgently. In this paper, a robust digital image watermarking algorithm, which embeds a meaningful binary image as the watermark, is proposed. Firstly, the watermarking algorithm scrambles the binary image used as the watermark by the toral automorphism system. Then, the discrete cosine transform on the blocks of a gay scale image used as the carrier image are carried out and the quantization table depicted in the JPEG standards, which are used when an image needs a relatively high compression rate, is used to quantize the DCT coefficients. After the quantization, some statistical characteristics of the coefficients which are still not equal to zero will be used to implement the embedding of the watermark. Given the embedding course of watermark mentioned above, this watermarking algorithm is expected to have great robustness, especially to JPEG compressing. The simulation experiment results also prove the effectiveness of the algorithm proposed.
Zhen Zhou, Shuyu Chen, Guiping Wang

Image Processing


The Performance Analysis of Low-Resolution Paintings for Computational Aesthetics

In the study of computational aesthetics, we always use high-resolution paintings to analyze painting style, but actually the paintings we obtain mostly are low-resolution. In this paper, the contrast experiments based on sparse coding are carried out between high and low resolution paintings. Different features are extracted in frequency domain and Gabor domain from the basis function of sparse coding (SC). Then the normalized mutual information (NMI) is figured out to analyze the effect of different features for painting style. At last, the features with better performance are used to classify the paintings’ style. The results of experiments show that, to a certain extent, the features extracted from low-resolution paintings still have the ability to characterize the painting style, among which the Gabor energy has the best effect in the painting style analysis.
Juan Zhu, Yuan yuan Pu, Dan Xu, Wen hua Qian, Li qing Wang

Using Mutual Information for Exploring Optimal Light Source Placements

Exploring optimal light sources for effectively rendering 3D scenes has been an important research theme especially in the application to computer graphics and visualization problems. Although conventional techniques provide visually plausible solutions to this problem, they did not seek meaningful correlations between proper light sources and their related attribute values. This paper presents an approach to exploring optimal light source placements by taking into account its correlations with such attribute values. Our idea lies in the novel combination of existing formulations by taking advantage of information theory. We first employ the quantized intensity level as the first attribute value together with the conventional illumination entropy so as to find the best light placement as that having the maximum mutual information. Meaningful relationships with viewpoints as the second attribute value are then studied by constructing a joint histogram of the rendered scenes, which is the quantized version of a 3D volume composed by the screen space and intensity levels. The feasibility of the proposed formulation is demonstrated through several experimental results together with simulation of illumination environments in a virtual spacecraft mission.
Yuki Ohtaka, Shigeo Takahashi, Hsiang-Yun Wu, Naoya Ohta

Extracting Important Routes from Illustration Maps Using Kernel Density Estimation

Illustration maps often direct our visual attention to the specific route with geographic symbols and annotation labels associated with important landmarks. This inspires us to evaluate the quality of such maps by analyzing the spatial distribution of visual attention over the map domain. In this paper, we introduce kernel density estimation in order to identify important routes that are implicitly designated by the map designers. Our algorithm begins by composing the density field as a combination of Gaussian kernels centered on the landmarks. The algorithm then allows us to extract an important route on the map as the trajectory of a ball running along the valley of the density field. We conducted a user study where we compared the routes reconstructed from the sequence of landmarks specified by the participants and their originally intended routes, and report some insight into possible aesthetic criteria in illustrating such maps.
Fumiya Sato, Hsiang-Yun Wu, Shigeo Takahashi, Masatoshi Arikawa

Fast Fractal Image Encoding Algorithm Based on Coefficient of Variation Feature

In order to improve the drawback of fractal image encoding with full search typically requires a very long runtime. This paper thus proposed an effective algorithm to replace algorithm with full search, which is mainly based on newly-defined coefficient of variation feature of image block. During the search process, the coefficient of variation feature is utilized to confine efficiently the search space to the vicinity of the domain block having the closest coefficient of variation feature to the input range block being encoded, aiming at reducing the searching scope of similarity matching to accelerate the encoding process. Simulation results of three standard test images show that the proposed scheme averagely obtain the speedup of 4.67 times or so by reducing the searching scope of best-matched block, while can obtain the little lower quality of the decoded images against the full search algorithm. Moreover, it is better than the moment of inertia algorithm.
Gao-ping Li, Shan-shan Li

A GNC Method for Nonconvex Nonsmooth Image Restoration

The augmented Lagrangian duality method have superior restoration performance for nonconvex nonsmooth images. However, an effective initial value could not be obtained for the augmented Lagrangian duality when it is used alone. To overcome this drawback, a hybrid method based on the augmented Lagrangian duality method and the graduated nonconvex method(GNC) is proposed. The better restored performance of the proposed method are illustrated by some numerical results.
Xiao-Guang Liu, Qiu-fang Xue

Posters and Demo Session


Research of Image Preprocessing in the Recognition System of RMB Crown Word

A crown word on a Chinese RMB note is a sequence of characters that uniquely identify a bank note. Training set sampling is very important in the recognition system of RMB crown word, which directly affects the accuracy of the recognition system. Image preprocessing operations require a number of properties on the crown characters such as the location, tilt correction, attitude, adjustment, noise reduction, binarization and single character segmentation and so on.
GuangJian Zhang

SPIDAR-S: A Haptic Interface for Mobile Devices

In order to introduce the haptic technology to more common users outside the academic sphere, this paper proposed a haptic device SPIDAR-S which is able to be attached to mobile devices such as smartphones and tablet PCs. The proposed device provides a user with one degree of freedom force feedback in the way of pulling one of the user’s index fingers with a string which is driven by a micro motor. Audio signals are adapted as the control signal for the motor’s output torques. Because of its simple structure and compacted size, SPIDAR-S is supposed to be taken around as a portable accessory for mobile devices, enabling a large number of users to enjoy haptic implemented applications in an easier and more casual way.
Shuhan Ma, Motonori Toshima, Kenji Honda, Katsuhito Akahane, Makoto Sato

An Interactive Haptization System in Video Contents Using SPIDAR-mouse

We propose a video haptization system. Several haptic devices have been proposed and are commercially available. However, they are still high price and are used for the researches mainly. In this circumstance, we had proposed the haptic device called Spidar-mouse that can easily construct the interactive system. In this paper, we developed the video haptization system with Spidar-mouse by using the movement information of the object in the video.
Kenji Honda, Ma Shuhan, Katsuhito Akahane, Makoto Sato

DroneMyo: Proactive Control of Unmanned Aerial Vehicle Based on Wearable Devices

Wearable devices promise a full immersion in the usage environment and most of these devices are mainly passive information receivers. We believe that integration of proactive wearable devices in general can bring more advantages for more natural user interactions. This paper explores proactive control of Unmanned Aerial Vehicle (UAV) based on a smart phone and a smart wristband. We have conducted indoor and outdoor trial flights and gained encouraging results as well as valuable ideas for future research.
Zhe Li, Yaxi Chen, Wenrong Tan


Weitere Informationen