Posted on Thu, April 11. The project must be submitted to the Teaching Assistant by Thursday April  25. The project must be submitted by electronic mail from your class account ya426XX to the class account ya42601@umd5.umd.edu. You must submit a readme file, your source files, and processed images. You must tar your multiple files into a single file named p2.tar.

Note: Graduate students will be working with  Dr. Fermüller  in the development of a web page concerning geometrically optical illusions due to bias. Basically, the home page will compile information about illusions and its corresponding explanations. Further, they will implement Java applets to illustrate some interactions with optical illusions. Undergraduate students that volunteer to work on this project should have permission granted by Dr. Fermuller. Therefore, if you are an undergraduate student and have no permission to work on the home page, you must to follow the description bellow. Otherwise, make an appointment with Dr. Fermuller.

Overview.

The goal of this project is to compute the flow field among the provided images (scene). Initially, you should smooth the images applying a Gaussian filter. The derivatives Ex, Ey and Et are computed to obtain the optical flow. The derivative Et is the difference between the two images. The derivatives Ex and Ey are computed taking the convolution of the two images with difference operators as Sobel (as in Project 1). The optical flow field is computed according to the optical constraint Ex u + Ey v + Et = 0. Since the optical constraint defines a line for each pixel in the image, you may compute the optical flow as the point (u, v) in this line that minimizes the distance to the lines corresponding to the neighbor pixels. However, you must choose one of the three following approaches to compute the optical flow field using any programming environment available in the OIT cluster, including MATLAB.

• B. D. Lucas and T. Kanade. An iterative image registration technique with an application to stereo vision. In Proc Int Joint Conf on Artificial Intelligence, pages 674-679, 1981.   Download

See also: Barron J.L., Fleet D.J. and Beauchemin S.S., "Performance of Optical Flow Techniques", International Journal of Computer Vision, Vol. 12, No. 1, pp. 43-77, 1994.   Download
• B.K.P. Horn and B.G. Schunck. Determining optical flow. Artificial Intelligence, 17:185-203, 1981.
• M. Proesmans, L. Van Gool, E. Pauwels and A. Oosterlinck. Determination of optical flow and its Discontinuities using non-linear diffusion. In 3rd European Conference on Computer Vision, ECCV'94, Volume 2, pages 295-304, 1994.

You must submit a readme file, named readme.txt. The readme file enumerates and describes all files in you tar file. The readme file should also contain a description of your implementations. You must also submit your source files and processed images.  Your may use MATLAB's quiver function to output an image with an edgel associated to each pixel (see figure below). All multiple files must be in a single tar file (p2.tar) and sent to ya42601@umd5.umd.edu.

Reference

• Robot Vision. B. Horn.
• Chapter 12 - Motion Field & Optical Flow.