Kyoungju Park

Deformable strokes towards temporally coherent video painting

We present an automatic and robust technique for creating non-photorealistic rendering (NPR) and animation, starting from a video that depicts the shape details and follows the motion of underlying objects. We generate NPR from the initial frame of the source video using a greedy algorithm for stroke placements and models, in combination with a saliency map and a flow-guided difference-of-Gaussian filter. Our stroke model uses a set of triangles whose vertices are particles and whose edges are springs. Using a physics-based framework, the generated and rendered strokes are translated, rotated and deformed by forces exerted from the sequential frames. External forces acting on strokes are calculated according to temporally and spatially smoothed per-pixel optical flow vectors. After simulating each frame, we delete unnecessary strokes and add new strokes for disappearing and appearing objects, but only if necessary to avoid popping and scintillation. Our framework automatically generates the coherent animation of rendered strokes, preserving the appearance details and animating strokes along with the underlying objects. This had been difficult to achieve with previous user-guided methods and automatic but limited transformations methods.

Motion estimation during photographing towards minimal solution for mobile panoramic images

Panoramic images, which are often produced off-line a while after the photos are taken, might be an entertaining application if they were generated on mobile devices immediately. Since producing a panorama requires feature extraction and matching that are expensive for mobile devices, an embedded panoramic system needs different approaches. Therefore, for on-line mobile panoramas, we propose methods that estimate camera motion during photographing and infer alignment based on motion so as to minimize computation time. A semiautomatic photographing method uses a guided user interface by showing a projected plane of the previously taken image and forces users to move the camera to the desired location. An automatic method keeps tracking the camera while the camera moves, estimates infinitesimal camera motions between neighboring frames, accumulates motion fields, and therefore infers correspondences of faraway frames. The resulting motion information can be used for initializing, aligning, and stitching to produce a panorama on a pre-computed common plane. However, incorrect pivoting, lens fall-off, or hand trembling may cause errors in stitching. So we apply a greedy local search algorithm to correct these errors. The complexity of the whole system is linear to the size of the images, and the running time is about 7 sec. for photographing and stitching four QVGA images on mobile devices.

Mutiple-view object tracking using metadata

In this paper we present a moving object tracking method in a multiple camera tracking system using meta-data. The proposed method acquires image sequences from multiple fixed cameras without a priori camera calibration. The tracking is performed based on objects robust characteristics, such as motion, shape, color, and etc. Elements of meta-data and the optimal tracking method are selected at each frame. Because objects characteristics are affected by camera position and environmental condition, tracking method between individual cameras is selected by optimal metadata. Tracking in a single view can be divided into three classes, such as; optical flow-based moving object detector, active shape model (ASM), color correlation based detector. The proposed method matches objects to most similar object color information, by comparing meta-data and ASM. In the experiment, the proposed multi-camera tracking system is show to be robust for non-rigid objects.

Enhanced player interaction using motion controllers for VR FPS

Goal of virtual reality (VR) is to increase the realism and user experience. In order to achieve these goals, we analyze the existing solutions and propose a new method of user interaction for a first person shooting (FPS) game prototype. In our FPS prototype, wearing a head mounted display together with two motion controllers, we design the player interaction using both hands to operate in-game firearms and mimicking shooting action in real-world. Additionally, we make use of one of the trending and free game engines in order to develop a run-able game prototype.

A Physical Interactive Game for Learning English of Children

As English is the universal language used by international society with immense interest, we ought to seek for more effective teaching method. One way to reach such goal is to utilize a physical interactive game in teaching English. A physical interactive game encourages the spontaneous activity in learning without the psychological burden while promoting interest in English and basic usage of English. Also, physical movement interface can stimulate learners’ motivation while inducing positive response in interest and immersion.

Pencil drawing animation from a video

We present an automatic, efficient, and simple technique to create pencil drawing animation, starting from a video. We generate pencil drawing from a source frame based on stroke modeling, specifying the properties of strokes, in combination with layered lines, flow‐guided difference‐of‐Gaussian (DoG) filter to several layers. Generated pencil strokes are translated and rotated because of forces exerted from the sequential frames using rigid body dynamics. Linear and angular forces acting on strokes are calculated according to the temporally filtered per‐pixel optical flow vectors. Our framework effectively generates the coherent animation of pencil strokes preserving the structured appearance of charcoal or pastel, which is difficult to achieve with previous‐abstraction based non‐photorealistic animation. Moreover, our stroke simulation step is suitable for animating the different styles of strokes, such as oil painting and watercolor, and can be efficiently implemented to produce animation in relatively inexpensive manner. Copyright

Visual analysis of soccer players and a team

Soccer is one of the most entertaining and popular sports around the world, and is also very interesting from a scientific point of view. Most of the scientific research on soccer is related to matches and game play analysis. In this paper, we propose a novel system for team performance analysis and visualization in terms of the structure of a team, and concentrate on cause-and-effect relationships between players and their teams based on player transfer data. Our system visualizes the individual player performance, team characteristics, comparisons between teams, and time varying changes of the team characteristics. The analyzed data are presented in two different ways (1) the system creates a pixel-grid visualization that presents the distinct characteristics of each player in a team. (2) A horizon graph is used to display the changes in team characteristics over time due to player transfers. This approach facilitates understanding the influence of player transfers on team characteristics in a very simple and straightforward manner.

Giant soap bubble creation model

We present a simple and real‐time technique for animating giant realistic looking soap bubbles. Unlike small spherical soap bubbles, giant soap bubbles stretch significantly and undergo globally varying and locally consistent deformations because of time‐varying surface tensions. For physically plausible surface tensions, we introduce and combine a variable‐length mass spring system and Gibbs elasticity and Marangoni elasticity. Using a proposed framework, we deform our models because of general forces from surface tensions, winds, and excessive pressures. Our framework effectively generates a physically plausible animation of giant soap bubbles, differing in shape according to the users design; generates capillary waves and vibrations while floating in the air; and shows vivid iridescent colors corresponding to the deformations, all of which have been difficult to achieve using previous soap bubble animation methods. Copyright

A serious game design for english education on Smart TV platform

In this paper we present a serious game design for HTML5 multi cross-platform environment. Previously existing application were restricted to their specific platforms and users could not enjoy them elsewhere (Smart phones, pc, Smart TV). In this paper we take advantage of the HTML5 multi cross-platform and make use of Smart TV technology in order to bring a new type of content to the big screen where more than one person can be involved, learn and enjoy.

Hybrid interface of a two-dimensional cubic Hermite curve oversketch and a three-dimensional spatial oversketch for the conceptual body design of a car

A spatial modelling system with a new interface which is optimized for automotive design is proposed in the conceptual modelling phase. In the initial sketch stage to set the volume and the ratio of the car, the automotive styling is created using a two-dimensional plane sketch with an accurate input and a designer-friendly interface. The character lines, which have important implications in the design of the car, are drawn with a curve-based oversketch. The shape of the car is modified using area-based modification. A completed two-dimensional model consists of converting the character lines to a three-dimensional car model. In the three-dimensional spatial oversketch stage, the edge and mesh on the surface are deformed by viewing the three-dimensional shape of the vehicle with various views using a spatial input device and stereoscopic display. Edge-based deformation modifies single or multiple edges while maintaining continuity between the surfaces. A sculpting technique is applied to execute special shapes such as the accent lines and the strake lines. A sketch system which can be used by designers to perform three-dimensional spatial modelling is proposed.