DOC PREVIEW
U of U CS 7960 - Lecture Notes

This preview shows page 1-2-21-22 out of 22 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 22 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 22 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 22 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 22 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 22 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

IntroductionCS7960 Advanced Image ProcessingWho should attend?PrerequisitesGoal and ObjectivesGoal and Objectives ctd.Learning ApproachGradingContentsContents IIContents IIIContents IVLinear DiffusionLinear vs. Nonlinear DiffusionNonlinear DiffusionShape Analysis & ModellingModelling ShapeHand ModelSkeletonSlide Number 20Slide Number 21Use of deformable models in VisionIntroductionGuido GerigCS 7960, Spring 2010CS7960 Advanced Image ProcessingIn-depth study of advanced methods and research topics of current interest in image processing and analysis. Covers PDEs, shape representations, deformable models (snakes, level sets), statistical shape analysis, scale-space and registration. Focus and list of topics might change from semester to semester.Who should attend?Graduate students who are interested in getting advanced knowledge in image processing and analysis, following the course 6640 Introduction to Digital Image Processing or equivalent. The course is particularly important for students involved in image processing research and will be a core course of the newly established imaging track.PrerequisitesCourse 6640 Introduction to Digital Image Processing or equivalent. Advanced programming skills related to imaging (C++, imaging libraries, Matlab or equivalent). Students with different background and curriculum need to discuss suitability and options with the teacher.Goal and Objectives• To discuss advanced topics in image processing and analysis that build on the introduction course.• Scientific methodology: Reading of scientific publications and book chapters, summarizing the contents, developing strategies to implement algorithms, and finally presenting the theory, tests, and applications to the audience.• To enable participants to implement solutions for complex image processing problems.Goal and Objectives ctd.• To enable participants to better understand novel, advanced methodology that is discussed in the image processing and image analysis literature.• To enable participants to teach image processing materials to the group by preparing and presenting a class lecture.Learning Approach• Students will read the relevant publications or chapters of books books and/or reading assignments BEFORE the class lectures.• In the course, the material will then be discussed in detail and motivated with real world examples and applications.• Students will actively participate in teaching by preparing a class lecture.• 5-6 projects where material discussed in class will be implemented, tested, and applied to image data and presented.Grading• Projects (5-6): 80%• Class Lecture: 10%• Participation: very important• Presentation of one project: 10%Late Policy: There will be 10 point deduction per week (out of 100) for late project submission.Contents• Feature Detection and Characterization– Notion of Scale Space– Gaussian Derivatives– Differential Invariant Structure– Nonlinear Scale Space– Anisotropic Diffusion– Diffusion of Higher Order DerivativesContents II• Shape Analysis– Fundamentals in Shape Analysis– Moment Invariants– Contour-based Invariants• Active Shape Models (ASM)• Active Appearance Models (AAM)• Elliptical Harmonics– Medial Axis RepresentationContents III• Object Segmentation– Generalized Hough Transform– 3D Deformable Models– Snakes– Level set evolution• Others– Normalized Graph Cuts– Etc.Contents IV• Image RegistrationLinear Diffusion• Edge locations not preserved• Region boundaries are preserved• Gaussian blurring is local averaging operation and does not respect natural boundariesSource: http://www.csee.wvu.edu/~tmcgraw/cs593spring2006/index.htmlLinear vs. Nonlinear Diffusion• Linear Diffusion: blur in all directions• Inhomogeneous diffusion: blur in all directions, but attenuate diffusion by an amount proportional to the magnitude of the gradient• Anisotropic diffusion: blur only in directions where the magnitude of the gradient is changing minimally, halt blurring across boundariesNonlinear DiffusionMultiscale image representation: Details in images should only exist over certain ranges of scale.Source: http://www.csee.wvu.edu/~tmcgraw/cs593spring2006/index.htmlShape Analysis & ModellingFrom: PhD thesis Rhodri DaviesModelling Shape• Define each example using points• Each (aligned) example is a vector123456xi= {xi1 , yi1, xi2 , yi2…xin , yin}Hand Model• Modes of shape variationb1b2bSkeleton• The Medial-Axis Transform (MAT) (Blum)is a specific skeletonisation, in which the greylevel of a skeleton pixel indicates distance to the original shape boundary.ShapeDistance T.Segmentation by deformable modelsFig. 1: Visualization of 3 MRI mid-hemispheric slices and the final positions (in red) of the automatic corpus callosum segmentation algorithm using deformable shape models.shape constrained Fourier snake 2D (CC-seg-avs1.mov)Deformable Models: SNAKESGeodesic Snake:Implementation: Curve C embedded as zero-level of higher order function ϕ[]Ntcακ+=∂∂() () ():)()(21ϕϕαϕϕϕϕϕ∇⋅+∇+∇⋅∇+∇⎟⎟⎠⎞⎜⎜⎝⎛∇∇⋅∇=∂∂∇+scgMCFrsrrrgcggggtUse of deformable models in


View Full Document
Download Lecture Notes
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Lecture Notes and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Lecture Notes 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?