EC-803 Computer Vision: Lecture-1
EC-803 Computer Vision: Lecture-1
EC-803 Computer Vision: Lecture-1
Lecture-1:
Course Introduction Basic Transformations- Translation, Scaling and Rotation, both in 2D & 3D Inverse Transformations Perspective Transformation and Imaging Process Relationship between Cartesian Coordinate System and Homogeneous Coordinate System Inverse Perspective Transformation
Course Introduction
Instructor: Mahmood Akhtar, PhD Lecture Timing: Mon 17302030 hrs, CR (Mts) -02 (Trg Block- III)
Topics:
Basic Transformations, Camera Model and Imaging Geometry, Camera Calibration and Stereo Imaging, Linear Filters, Multiview Geometry, Structure From Motion, Segmentation, Finding Images in Digital Libraries, Tracking, Model-Based Vision, Finding Templates Using Classifiers, Range Data
Prerequisites:
Digital image processing Working knowledge of Matlab Knowledge related to: Euclidean and projective geometry Linear Algebra Vector calculus Probability & Statistics
Yahoo Group:
CV_CEME
Grading Policy*:
Surprise quizzes Programming assignments Sessional exam I Sessional exam II Project Final exam
*Relative final grading policy
Projects:
Projects will carry 15% weight in the total marks Projects may be conducted individually or in groups of max. two students Suggested topics for projects will be uploaded on the class yahoo group some time between week 2 and week 4 of the course Project topics should be selected and approved within the first six weeks of the course. Project presentations will commence from week 12 onwards and projects (report, CD & presentation) will not be accepted after the presentation date. Projects consisting of downloaded codes or presentations will not be accepted and will carry ZERO MARKS
Vision
Process of discovering what is present in the world and where it is by looking
Computer Vision
goal is to emulate human vision (which is limited to the visual band of electromagnetic (EM) spectrum), including learning and being able to make inferences and take actions based on visual inputs
Related Disciplines
Image processing Pattern recognition Computer graphics Artificial intelligence Machine learning
Basic Transformations
Translation (2D)
Basic Transformations
Scaling (2D)
Basic Transformations
Rotation (2D) - around origin
**consult your class notes for further information on this topic**
Basic Transformations
Rotation (2D)
- around any other point (not origin) T-r(R T-rp)
1. take the translation to origin 2. rotate an angle required 3. take inverse translation
Basic Transformations
Translation (3D)
Basic Transformations
Scaling (3D)
Basic Transformations
Rotation (3D) - around z axis
**consult your class notes for further information on this topic**
Basic Transformations
Rotation (3D)
- around x axis
**consult your class notes for further information on this topic**
Basic Transformations
Rotation (3D)
- around y axis
**consult your class notes for further information on this topic**
Basic Transformations
Concatenation - several transformations can be presented by a single
4x4 transformation matrix V* = R (S (TV)) = AV - matrix operations are not commutative
Basic Transformations
Transformation for a Set of Points
- for a set of m points, construct a matrix V of dimension 4 x m, such that each column represents a point - transformation: V* = AV - ith column, jth element of V* is transformed point corresponding to Vi,j
**consult your class notes for further information on this topic**
Inverse Transformations
Inverse translation (sign is changed)
Inverse Transformations
Inverse scaling (values get inverted)
Inverse Transformations
Inverse rotation (angle sign is changed)
Problem is that mapping of 3D scene onto image plane (2D) is many to one mapping (x0,y0) is the mapping of points on a straight line passing through (x0,y0,0) and (0,0,)
Inverse Perspective Transformation is formulated by using Z component of Ch as a free variable, which gives:
We can find out X and Y only and require additional information to find out Z so that point in3D world coordinates is exactly known from x0, y0 in image plane SO FAR: Both coordinate systems were aligned NEXT: Imaging geometry where world coordinate system and camera coordinate system are not aligned