课程名称 (Course Name) ： Computer Vision

课程代码 (Course Code)：   C032703/F032528

学分/学时 (Credits/Credit Hours)： 3/48

开课时间 (Course Term )：  Spring

开课学院（School Providing the Course）:  School of Electronic，Information and Electrical Engineering

任课教师（Teacher）:  Xu Zhao

课程讨论时数（Course Discussion Hours）:  4

课程实验数（Lab Hours）:   8

课程内容简介(Course Introduction):

Computer vision aims to recover useful information about a 3D scene from its 2D projections (images), such as the depth and structure, motion, surfaces curvature and orientation of 3D objects and status and meaning of the actions of 3D scene. In this course, basic concept, theories and algorithms of computer vision are introduced. First, how an image could be formed is systematically analyzed, from the viewpoints of geometric camera models, light, shading and digital camera. Then, the theories and algorithms about image filtering, binary image processing and local image features will be reviewed. Multiple-view based techniques as stereo vision, 3D motion analysis and image alignment are then studied.  Mid-level and high-level vision topics, such as optical flow, tracking, segmentation, object detection and recognition will be discussed in details.

教学大纲(Course Teaching Outline):

Part 0: Introduction

Course 1: Introduction

1)      Concept of computer vision

2)      Related fields

3)      Human vision

4)      State of the art

Part I: Image formation

Course 2: Geometric camera models

1)      Geometric primitives and transformations

2)      Projection models

3)      Intrinsic and extrinsic parameters

4)      Camera calibration

Course 3: Light and shading

1)      Lighting

2)      Reflectance and shading

3)      Shape from shading

Course 4: Digital camera

1)      Image sensing pipeline

2)      Sampling and aliasing

3)      Color

Part II: Early vision – single view

Course 5: Image filtering

1)      Linear filters and convolution

2)      Non-linear filters

3)      Spatial frequency and Fourier transforms

4)      Histogram equalization

5)      Image pyramids

Course 6: Binary image processing

1)      Binary image generation

2)      Binary image representation

3)      Morphological processing

Course 7: Local image features

1)      Image gradient

2)      Edges

3)      Lines

4)      Points and patch

5)      SIFT and HOG

6)      Texture

Part III: Early vision – multiple views

Course 8: Stereopsis

1)      Binocular camera geometry

2)      Epipolar constraint and fundamental matrix

3)      Stereo of arbitrary camera arrangement

4)      Structured Lighting

Course 9: Motion

1)      Motion from 3D PCs

2)      Motion from 2D PC

3)      Motion from LC’s

4)      Motion from other image clues

Course 10: Alignment and warping

1)      Background

2)      Mosaics and warping

3)      Outlier processing: RANSAC

Part IV: Mid-level vision

Course 11: Tracking

1)      Optical flow

2)      Linear dynamical models and Kalman filters

3)      Particle filtering

Course 11: Segmentation

1)      K-means and EM

2)      Graph cuts and energy-based methods

Course 12: Grouping and fitting

1)      Hough transform

2)      Deformable contours

Part V: High-level vision

Course 13: Object detection

1)      Sliding window method

2)      Pyramid match kernel

3)      Part-based models

Course 14: Recognition

1)      Generative categorization: Naïve Bayes model for classification

2)      Discriminative classifiers: SVM

3)      Bag-of-words models

课程进度计划(Course Schedule):

Week num.	Topic	Assignments
2	Intro.
3	Geometric camera models
4	Light and shading
5	Digital camera	Problem set 1 out
6	Image filtering
7	Binary image processing
8	Local image features
9	Stereopsis	Problem set 2 out
10	Motion
11	Alignment and warping
12	Tracking
13	Segmentation	Problem set 3 out
14	Grouping and fitting
15	Object detection and Recognition
16	Exam. review
17	Final exam.

课程考核要求(Course Assessment Requirements)：

Problem set:           40%

Final exam:             60%

参考文献(Course References)：

[1] Computer Vision: Algorithms and Applications. Richard Szeliski.

[2] Computer Vision: A Modern Approach. David A Forsyth

[3] Multiple View Geometry in Computer Vision. Richard Hartley et al

[4] Machine Vision. Ramerh Jian, et al

[5] Vision. David Marr.

[6] Pattern Recognition and Machine Learning. Bishop.

[7] 机器视觉,贾云得 编著

[8] 计算机视觉的理论和实践, 李介⾕谷著

[9] 计算机视觉——计算机理论与算法基础, ⻢马颂德 著

预修课程（Prerequisite Course）

(1)  Digital image processing

(2)  Mathematics: Matrix, Probability theory

(3)  Programming: C, C++, Matlab