SYDE 575: Image Processing Introduction Read Textbook Chapters 1 & 2 What is Digital Image Processing?
Definition: manipulation of digital image for enhancement, compression, transmission, information extraction or analysis manipulation involves denoising, enhancement, registration, color mapping, rescaling, feature point detection, illumination reduction, Digital image defined as a 2-d function: f(x,y) x and y are spatial (in-plane) coordinates
f(x,y) is the value (usually stored as discrete) at a spatial (x,y) location With multiple images of the same scene, one can deduce 3-d geometry and pixels in the image could have 3-d coordinates e.g., f(x,y,z) where (x,y,z) represents spatial coordinates A video is defined as a 3-d function: f(x,y,t) where t represents time
Computer Vision Scene Image Processing
Image Analysis Image Understanding Information
Image Analysis: content extraction e.g., segmentation, shape description, boundary detection, mathematical morphology, texture feature extraction, motion estimation Image Understanding: decision making based on content extraction; covered in SD372 and SD675 SD575 deals primarily with Image Processing and, later in
the course, Image Analysis Why is image processing difficult? Mapping from a 3-d world to a 2-d plane Measured intensity is a function of many factors
Interpreting groups of pixels as interesting objects is easy for the human, but not for the computer Loads of data to process! Local windows versus global interpretation Early Image Processing
Newspapers needed to send pictures across the Atlantic Ocean quickly Source: Gonzalez and Woods Early Image Processing Capturing and transmitting images from
space Source: Gonzalez and Woods Electromagnetic Spectrum Source: Gonzalez and Woods
Machine Vision Source: Gonzalez and Woods Thermal (Infrared) Imaging
http://www.nationalinfrared.com/image_browser.php Satellite VIR (Visible Infrared) Imaging Source: Gonzalez and Woods
Satellite Radar Image of Sea Ice Source: MDA X-Ray Imaging Source: Gonzalez and Woods
MRI Images Source: Gonzalez and Woods Ultrasound Imaging
Source: Gonzalez and Woods Grey Scale vs Color Grey scale: single band Color: three images (red, green, blue) combined to create single image
Image Sensing Source: Gonzalez and Woods Digital Image Acquisition Example
Source: Gonzalez and Woods Simple Image Formation Model Image may be characterized by:
Amount of source illumination incident on scene Amount of illumination reflected by objects in scene (r=0 for total absorption, and r=1 for total reflectance) f ( x , y ) = i ( x , y )r ( x , y ) 0 < i( x , y ) < 0 < r( x, y) < 1
Digital Image Representation Source: Gonzalez and Woods Image Sampling and Quantization
Source: Gonzalez and Woods Example Source: Gonzalez and Woods
Spatial Resolution Source: Gonzalez and Woods Gray-level Resolution Source: Gonzalez and Woods
Fundamental Steps Source: Gonzalez and Woods Vision and Image Processing (VIP) Lab
UW Research Lab that conducts research in computer vision Covers many applications: remote sensing, biomedical, video analytics, 3d reconstruction, etc. Many connections to industry to conduct applied research
Directors: Profs. Clausi, Wong, and Fieguth http://vip.uwaterloo.ca Graduate Studies If you are interested in graduate studies, chat with faculty members in your field of interest Make sure that you apply for scholarships
NSERC/OGS applications due typically in October We are always looking for a few new graduate students to conduct research in the VIP lab (start Spring or Fall terms)
