gpgpu present 1

12
698470025 薛健鑫 R98944017 蔡昆修 R98922041 何星翰

Upload: uc-davis

Post on 21-Jan-2015

559 views

Category:

Documents


5 download

DESCRIPTION

The first presentation of gpu project.

TRANSCRIPT

Page 1: Gpgpu present 1

698470025 薛健鑫

R98944017 蔡昆修

R98922041 何星翰

Page 2: Gpgpu present 1

Super-resolution (SR) Super-resolution is a techniques that enhance the

resolution of an imaging system.

Two approach of SR

single-frame approach

multiple-frame approach

Page 3: Gpgpu present 1

Super-resolution (SR)original low-resolution sequential frames of video

The same frames zoomed 4 times

Zoomed with super resolution

Page 4: Gpgpu present 1

Super-resolution (SR)

Page 5: Gpgpu present 1

Method (multiple-frame)

Step1. High Quality Spatial Upscaling(Lanczos3)

Step2. Motion CompensationAlgorithms

try to extract details from every image in a sequence to reconstruct other frames.

Page 6: Gpgpu present 1

Upsize Upsize all frames 2 times using interpolation

Using Lanzcos filter to get high quality result

Page 7: Gpgpu present 1

Motion compensation Motion compensation is a technique used by video

compression

Suppose the only difference between one frame and another is the result of either the camera moving or an object in the frame moving.

Motion compensation takes advantage of this to provide a way to create frames of a movie from a reference frame.

Page 8: Gpgpu present 1

Motion compensation In this case, we use motion compensation technique

for finding corresponding areas in neighbor frames

Page 9: Gpgpu present 1

User Interface: Video Player

Interactive Video Player

Can zoom in and zoom out when watching video

High Quality SR image when zoom-in

Page 10: Gpgpu present 1

Cuda acceleration

Video Encode and Decode

Upsize image , Lanczos3 interpolation

Motion compensation

Page 11: Gpgpu present 1

Real time system(final goal) Using CUDA to get real-time SR video player

At least fast off-line system

Page 12: Gpgpu present 1

Reference1. R. Y. Tsai and T. S. Huang, "Multiframe image restoration and registration," in Advances

in Computer Vision and Image Processing, vol. 1, chapter 7, pp. 317-339, JAI Press, Greenwich, Conn, USA, 1984.

2. M. Irani and S. Peleg. 1991, "Super Resolution From Image Sequences" ICPR, 2:115--120, June 1990.