Image Processing Techniques (CMSE890-001)

Image Processing Techniques (CMSE890-001)

January 8th, 9:00AM to May 4th, 10:00AM

Develop and explore tools that assist researchers in analyzing scientific image datasets. This course focuses on computational representation of images and types and classes of algorithms that have been developed for science analysis.

Event Categories:  Trainings

Course Time: MWF 9:00 AM -10:00 AM

Motivation: Analyzing vast amounts of image data for science continues to be a time consuming process. The reducing cost of gathering image-based data with commodity cameras (e.g., cell phone cameras) and imaging sensors (MRI, CAT Scans, Lidar, etc.) has created a "big data" problem in science. Analysis of images to extract scientific measurements  is still a time-consuming process, which is only accentuated by this avalanche of data. Central to the problem is that new scientific questions and image datasets require custom image analysis workflows. Even when domain scientists are aware of the types of algorithms/solutions available, it is expensive to develop custom software analysis tools for specific research problems or domains. For this reason, many researchers find it faster and cheaper to manually annotate and process images by hand using low-level tools such as Photoshop.

This course intends to develop and explore tools that assist researchers in analyzing their scientific image datasets. We focus on the computational representation of images and the types and classes of algorithms that have been developed for science analysis.

Topics covered include, but are not limited to:

  • Data representations
  • Methodologies for acquisition
  • Preprocessing
  • Binary morphology
  • Segmentation
  • Feature selection
  • Machine learning
  • Visualization
  • Computational techniques for dealing with big data (clusters, accelerators, etc.)

Prerequisites: Programming experience is expected (CMSE 801 or equivalent). Most of the course will be taught in Python using Jupyter notebooks. Prior knowledge of Python is not required, however, previous experience in a programming language is expected. Students will also be introduced to other imaging tools such as: ImageJ (Fiji), Matlab, OpenCV, ImageMagick, and FFMpeg.

Instructor: Dirk Colbry

Email: colbrydi@msu.edu

Office: 1516 Engineering Building

United States
US