Eric Mortensen, Assistant Professor of Computer Science, created the Magnetic Lasso Tool that people use every day in Adobe Photoshop. Learn more about Mortensen, his research, and why he's chosen this career.

Q: How did your work on the magnetic lasso come about, and how were you able to transfer the technology to Adobe?
A: The magnetic lasso in Photoshop is based on a similar tool called intelligent scissors that I developed for my Master's thesis. The idea for intelligent scissors started in January 1992. I was a first-year graduate student taking the Advanced Computer Vision course at BYU from Dr. William Barrett. In preparation for our first programming assignment (a user-assisted local edge follower), Dr. Barrett was detailing in class several local edge following techniques as well as some dynamic programming/graph searching methods. As he was describing one such technique, it occurred to us that, rather than try to guess how far apart to space the control points before computing the optimal path between them, a user could select each goal point after computing optimal paths to all other points in the image; displaying the optimal path from each cursor position back to the user specified control point (called a seed point) as the mouse is moved. Since I was involved in the discussion that led to this idea, I volunteered to try this idea for my first assignment in lieu of a local edge following approach. We didn't realize at the time that this seemingly simple idea would provide such a powerful and effective user feedback mechanism.

Well, I tried to implement a pixel-based, recursive version of the lattice DP algorithm but I couldn't get it to work. With the assignment deadline just one day away, I completely abandoned the algorithm in favor of what turned out to be a much more efficient implementation. When Dr. Barrett first saw the results (I had stayed up late the night before to finish the program), he described the interactive feedback as a "live-wire" because of the way it jumps and snaps to object boundaries in response to cursor movement.

Examples of the intelligent scissors at work, selecting specific objects in a photo. Due to the interactive optimal path selection inherent in Intelligent Scissors, the user is able with minimal guidance to define an object boundary.

In 1995, we presented our intelligent scissors work at SIGGRAPH and demonstrated its utility in image composition tasks. Paul Asente, from the Adobe Photoshop development team, was in the audience and became interested in adding Intelligent Scissor to Photoshop. After a few e-mails, a couple of visits to the Bay Area for a live demo and technical discussions about computational and memory requirements, and a lot of wrangling by lawyers (in which aspect I was happily out of the loop), Adobe finally ended up buying my Master thesis code, which they then incorporated into Photoshop as magnetic lasso.

Q: Tell us a bit about your experience and background before coming to OSU.
A: I graduated with a BS in 1991, an MS in 1995, and a PhD in 2000, all in Computer Science (with a Physics minor for my BS) and all from BYU. My general research interests are in image processing, computer vision, and computer graphics. As noted above, I developed intelligent scissors for my Master's thesis. My PhD work added several extensions to intelligent scissors to further simplify image and video editing. In addition to my work with intelligent scissors, I also worked on projects for image-space contour interpolation, color quantization with dithering, a couple of graphics related projects, and writing an object-oriented image processing library.

Q: Why did you choose to major in computer science?
A: My father was a high school math teacher so I always wanted to have a science/engineering type of career. During my teenage years, my father became a computer programmer and took me to work one day. He helped me write a very simple Fortran program that printed out designs using text. Because of that, I took the computer math class during my junior year of high school. It was the first year they offered it and many of the students learned to program (in BASIC on the RadioShack TRS 80) as fast as the teacher, if not faster. I moved to California for my senior year and continued to take computer-related classes, this time on Apple IIe computers, where I wrote simple 2-D and 3-D graphics programs. During this time, my best friend had bought a Commodore 64 and I started writing programs on it as well, including a maze generation program. Prior to entering college, my parents bought me a new 4.77 MHz 8088 PC (switchable to 8 MHz) with EGA graphics capabilities (which lasted me through all of my BS degree and some of my MS). I bought a book and Borland Turbo Pascal and taught myself to program in Pascal. When I started college, I decided to major in computer science because of my previous background (although I was also considering computer engineering).

Q: Why did you choose to come to work at OSU?
A: The faculty are very collegial. I like the area. I have relatives in Oregon. They made me a good offer.

Q: What do you hope to accomplish at OSU?
A: Tenure. :^) I want to be able to develop algorithms that solve (or at least make major advances in solving) some of the fundamental problems in computer vision such as segmentation, correspondence, and object recognition/classification.

Q: What's the most exciting thing about your research?
A: Most areas of computer science take advantage of a computer's ability to process numerical and textual data. A computer can calculate, search, sort, etc. much faster than a human. However, when it comes to processing visual/perceptual data, a computer's capabilities fall far short of what a person can do. As such, computer vision still has several challenging problems that need to be solved before a computer can "see".

Q: About your teaching?
A: The most satisfying aspect of teaching is when students understand concepts that they didn't previously understand.

Q: What other practical ("real-world") applications might your research have?
A: "Real-world" applications include image and video composition, medical image diagnosis, object tracking in video (for both military and commercial applications), and object recognition/classification.

Q: How would you describe OSU to someone who's never been here?
A: It's a good university that is getting better. The people and area are great.

Q: What do you like to do in your spare time (if you have any)?
A: Read and cook.