01_introduction - Computer Science and Computer Engineering

CSCE 5013
Computer Vision
Fall 2011
Prof. John Gauch
01 - Introduction
Application Areas
Course Objectives
• Computer vision is the process of extracting
useful information from digital images
– Finding objects of interest in images
– Properties of objects (size, shape, color)
– Recognition of objects
• Computer vision is also known as machine
vision, robot vision, computational vision, or
image understanding
• The fundamental problem of computer vision
is that multiple models could fit the image data
– Fitting a line equation to set of 2D points
– Calculating 3D coordinates from 2D images
• Hence we must select the best model that fits
the data given the time/space constraints of the
• Computer vision is closely related to three
other research areas:
Image processing (image => image)
Computer vision (image => model)
Computer graphics (model => image)
Computational geometry (model => model)
Application Areas
• Automated inspection – CV is used to look for
defects in manufactured parts and to assist in
automated assembly
Application Areas
• Navigation – CV is used to guide a car or robot
along roads or paths while avoiding obstacles
Application Areas
• Computer graphics modeling – CV is used to
generate natural looking models that bend and
move like real objects
Application Areas
• Security and surveillance – CV is used to
watch areas of interest to detect suspicious
activities in restricted areas
Application Areas
• Medical applications – CV is used to locate,
identify and quantify abnormal features in
medical images and assist in treatment
Application Areas
• Human Biometrics – CV is used to recognize
people via face or fingerprint recognition
• Computer vision is a well established area of
computer science and engineering
– First attempts to model blocks world images were
made at MIT in the 1960s
– In the 1970s early computer vision methods made
use of AI techniques to reason about line drawings
– In the 1980s attempts to “understand everything”
in an image were outperformed by task specific
– The focus in the 1990s moved towards more
physics based image modeling and real time
applications such as video content analysis
– In the 2000s we have seen computer vision
methods mature and become widespread in other
areas such as computer graphics
– In the 2010s we will see even wider use of
computer vision applications making use of
FPGAs and GPUs and mobile devices
Course Objectives
• Goal of this class is to learn the fundamental
techniques used in computer vision
Mathematical tools and techniques
Algorithms and data structures
Existing computer vision software
Developing CV applications in C++
Reading current research literature
Course Objectives
• The remainder of this class we will focus on
the fundamentals of computer vision
– Image formation – how digital images are captured
and how this knowledge of the scene can be used
– Image processing – survey of basic techniques to
manipulate images prior to detailed analysis
– Feature detection – methods to extract geometric,
chromatic or textural features from images
Course Objectives
– Image segmentation – methods to partition an
image into visually sensible regions
– Feature-based alignment – how image features can
be used to align large collections of images
– Motion estimation – techniques for measuring
camera and/or object motion in image sequences
– Emerging techniques – discussion of recent trends
in computer vision and future applications