• Journal of the Korea Computer Graphics Society
• /
• v.19 no.1
• /
• pp.11-20
• /
• 2013

This paper introduces a user interface for large 3D meshes in mobile systems, which have limited memory, screen size and battery power. A large 3D mesh is divided into partitions and simplified in multi-resolutions so a large file is transformed into a number of small data files and saved in a PC server. Only selected small files specified by the user are hierarchically transmitted to the mobile system for 3D browsing and rendering. A 3D preview in a pop-up shows a simplified mesh in the lowest resolution. The next step displays simplified meshes whose resolutions are automatically controlled by the user interactions. The last step is to render a set of detailed original partitions in a selected range. As a result, while minimizing using mobile system resources, our interface enables us to browse and display 3D meshes in mobile systems through real-time interactions. A mobile 3D viewer and a 3D app are also presented to show the utility of the proposed user interface.

• Journal of the Korea Computer Graphics Society
• /
• v.14 no.1
• /
• pp.27-33
• /
• 2008

Mobile systems have relatively limited resources such as low memory, slow CPU, or low power comparing to desktop systems. In this paper, we present a new 3D mesh connectivity coding algorithm especially optimized for mobile systems(i.e., mobile phones). By using adaptive octree data structure for vertex positions, a new distance-based connectivity coding is proposed. Our algorithm uses fixed point arithmetic and minimizes dynamic memory allocation, appropriate for mobile systems. We also demonstrate test data to show the utility of our mobile 3D mesh codec.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2022.11a
• /
• pp.42-43
• /
• 2022

본 논문은 모바일 단말에서 실시간으로 동작하는 딥러닝 기반 경량 초고해상화 기술에 관한 내용이다. 대용량 3차원 메쉬 모델의 비실시간 압축은 실시간 스트리밍 응용 시나리오에서 제약점으로 작용하고 있고, 본 논문에서는 두 입력 텐서의 차원을 일치시켜야 하는 element-wise 덧셈 대신 concatenation을 활용해 연산량을 개선하고, float-to-int8 양자화 과정에서 발생하는 오차를 줄이기 위해 weight clipping 및 regularization 기법을 활용해 초고해상화 화질 성능을 개선하였다. 제안하는 알고리즘은 기존 모바일 초고해상화 기술을 화질 측면에서 0.12dB, 처리 속도 측면에서 13.6ms 개선하였고, Mobile AI & AIM 2022 실시간 이미지 초고해상화 대회에서 1등을 수상하였다.