Instructor: 劉 興 民 
Head Teaching Assistant 首席助教
Lectures: Tue 9:10-12:00 王偉曛 asdfg3215740@gmail.com
工學院A館205教室 工學院A館410室
Office: 工學院A館403室 分機23153
e-Mail: damon@computer.org
Phone: 分機33118
Computational photography is an emerging field at the intersection of
computer graphics, image processing, machine vision, applied optics,
and increasingly, machine learning.
This class will look into how computation can be used to capture,
create, and enhance digital imagery in an effort to move beyond the
constraints of traditional cameras.
This course covers fundamentals and applications of hardware and
software techniques, with an emphasis on software methods.
Topics include cameras and image formation, image processing and
image representations, high-dynamic-range-imaging, human visual
perception and color, single-view and multiple-view 3D model
reconstruction, morphing,
data-rich photography, super-resolution, and image-based rendering.
The course will explain the principles behind many of the advanced
tools that can be found in Adobe Photoshop, although the use of this
package itself is outside the scope of the course.
Students will have written programs to create optical illusions,
add or remove objects from a photograph, smoothly morph
between faces, automatically stitch together photos into panoramas,
and more.
隨著數位相機之進步與普及，計算攝影學已成為電腦圖學之一重要研究方向。
計算攝影學研究如何利用數位計算來結合、編輯及改善數位影像、
突破一般相機的影像捕捉限制及有效率地利用大量的高解析度數位相片等課題，
如利用全景相片或 super-resolution 提升傳統相機之解析度，高動態範圍影像
(high dynamic range imaging) 能更精確地記錄場景的能量 (radiance)、
改善相機之光學元件及感測器以記錄多視角之影像 (light field)
以達到動態重聚焦 (re-focusing) 的效果，
結合使用閃光燈及不使用閃光燈的影像提升成像品質，
開發新演算法去除由於震動 (shake) 造成的影像模糊 (blur) 等。
這些技術隨著數位相機的普及將有非常廣泛的應用，
使計算攝影學研究同時兼具高度之學術及實用價值。
〔節錄自國科會資訊工程學門研究領域規畫書內容〕
Readings
Indicative Topics
- A crash course in OpenGL
- Cameras and image formation
- Color and human visual perception
- Aesthetic photo quality assessment
- Interactive image segmentation
- Gradient image manipulation
- Matting & inpainting
- Texture synthesis & filling
- Warping & morphing
- Structure aware retargeting
- Panorama stitching
|
- Internet-scale photo collections
- Photo tourism and structure from motion
- High dynamic range images
- Tone mapping and bilateral filters
- Image restoration and motion deblurring
- Optics & lens technology
- Defocus and depth of field
- Computational illumination
- Light field capture & rendering
- GPU data-parallel computing
- An introduction to deep learning
|
Assessment
Laboratories [25%];
Midterm [33%];
Final exam [20%];
Final project / PowerPoint slides [12%];
Punctuality, attendance, and
participation in discussions [10%].
These weights are tentative
and subject to future adjustment.
為鼓勵大學部同學修習本科目，對其將採取不同的評分標準：
1) 大學部同學的修課人數將獨立，不納入修課總人數來計算；
2) 大學部同學的學期個人總成績，將依據其他研究所同學調整之後的分數，
向上再提昇至少2至5分 (但以99分為上限)。
Course Works
課業倉儲
Discussion Board
課程討論區