Version 1.29: Faster rendering.
This page is a companion to:
IEEE Transactions in Medical Imaging , M.D. Abràmoff, W.J. Niessen and M.A. Viergever: Objective Quantification of the Motion of Soft Tissues. IEEE TMI. 2000; 19 (10): 986-995.
FlowJ is a collection of popular 2D optical flow algorithms, Lucas and Kanade, Uras, Fleet and Jepson, and Singh, in Java. It has been interfaced to ImageJ, the Java image processing program developed by Wayne Rasband at the National Institutes of Health. If properly installed, FlowJ appears as an extra command in ImageJ under the Plugins menu. It can measure and visualize the image flow (related to motion) in any open stack such as new_square2.tif to the right. The test image sequence is available at ftp.csd.uwo.ca/pub/vision).
The images under it show the flow field in different formats. Formats available in FlowJ include the DC format (a pseudocolor format developed by myself), the common quiver (also called arrow) format (detail only because it requires enlargement) and spotnoise (a thermal flow rendering format developed by J.J. van Wijk). The flow field is always shown as a grayscale (float) stack with separated x and y flows.
One day, really, there will be a FAQ.
DCM (Dynamic Color Map)
I release this implementation in the public domain provided that anyone who uses them in research and publishes anything about it reference me:
M.D. Abràmoff, W.J. Niessen, and M.A. Viergever. “Objective quantification of the motion of soft tissues: an application to orbital soft tissue motion”. IEEE Trans Med Imag. 19 (10):986-995, 2000.
I do not allow commercial use.
Michael Abràmoff. (personal data)
Unzip the files into the ImageJ directory or folder. In WinZip, check "Use folder names".
Be sure that you end up (at the ImageJ level) with a plugins folder containing the bij.jar and FlowJ (and also all other parts of BIJ) directory or folder. Previous versions made directories linearalgebra, bijnum or of2d, which can be safely removed.
To start, run ImageJ. Open a suitable grayscale 8-bit stack. Under the Plugins menu, you will see a new command:
FlowJ->FlowJ. If you run it you will see a dialog with the following options:
The parameters refer to the differing algorithms (LK = Lucas and Kanade, U = Uras, FJ = Fleet and Jepson, S = Singh). The parameters not mentioned above are very specific to each algorithm. The best explanation for these is found in my TMI or the Barron et al. paper.
back to top
Recently discovered the Carnegie Mellon University image database with many good motion sequences, available at http://vasc.ri.cmu.edu/idb/html/motion/index.html
back to top
- Barron, J. L., Fleet D.J., and Beauchemin S.S., "Performance of Optical Flow Techniques," Int J Comp Vis, vol. 12, no. 1, pp. 43-77, 1994. The best reference overall.
- Niessen, W. J., Duncan, J. S., Nielsen, M., Florack, L. M. J., and ter Haar Romeny, B. M., "A Multiscale Approach to Image Sequence Analysis," Comp Vis Imag Understand, vol. 65, no. 2, pp. 259-268, 1997.
- Lucas, B. and Kanade T., " An Iterative Image Registration Technique with an Application to Stereo Vision", in Proc. DARPA Image Understanding Workshop, 1981.
- Fleet, D. J. and Jepson, A. D., "Hierarchical Construction of Orientation and Velocity Selective Filters," IEEE Trans Patt Anal Mach Intell, vol. 11, no. 3, pp. 315-325, 1989.
- Fleet, D. J. and Jepson, A. D., "Computation of Component Image Velocity From Local Phase Information," Int J Comp Vis, vol. 5, no. 1, pp. 77-104, 1990.
- Uras, S., Girosi, F., Verri, A., and Torre, V., "A Computational Approach to Motion Perception," Biol Cybern, vol. 60 pp. 79-87, 1988.
- Singh, A., " An estimation-theoretic framework for image-flow computation", in Proc. 3rd Intern. Conf Comput Vis., 1990.
- M.D. Abràmoff, W.J. Niessen and M.A. Viergever: Objective Quantification of the Motion of Soft Tissues. IEEE TMI. 2000; 19 (10): 986-995.
- Wijk, J.J. van. Spot noise - Texture Synthesis for Data Visualization.
Computer Graphics, 25(4), (Proceedings SIGGRAPH'91), 1991, 309-318.
If you have ftp, look at ftp.csd.uwo.ca/pub/vision (anonymous login with your password) to find the C-sources for all optical flow algorithms that accompanied the Barron et al. article and that were a major source of inspiration for me. You can also find example image sequences to test with there.
(c) 1999-2003, Michael Abramoff. Last updated 2003/12/7.