Image Sequence Analysis: Motion Estimation
Military problems — Tracking of multi targets from video data. Measuring missile dynamics from video data. Target detection and recognition in Forward Looking Infrared (FLIR) image sequences.
Industrial problems — Dynamic monitoring of industrial processes. Dynamic robot vision.
Commercial problems –– Bandwidth compression of TV conferencing and picture phone video signals.
Medical problems — Study of cell motion by mi crocinematography. Study of heart motion from X-ray movies.
Meteorology — Cloud tracking.
Transportation — Highway traffic monitoring.
Image sequence processing involves a large amount of data. However, because of the rapid progress in computer, LSI, and VLSI [1.1] technologies, we have now reached a stage when many useful processing tasks for image sequences can be done in a reasonable amount of time.
One of the most important issues in image sequence processing is motion estimation. In many image sequence processing problems, motion estimation is the key issue. For example, in efficient coding using DPCM in time, motion estimation and compensation can potentially improve the efficiency significantly. In reducing noise in image sequences by temporal filtering, registration of the object of interest from frame to frame is necessary, and registration is, in essence, equivalent to motion estimation. Finally, in tracking multiple targets (moving differently), motion estimation provides a powerful way of segmenting and identifying the individual targets.
Because of the importance of motion estimation, it will be the main concern of the present chapter. After a brief outline of the contents of the book in the next section, the remainder of this chapter is devoted to a discussion of motion estimation techniques.
Unable to display preview. Download preview PDF.
- 1.1D.F. Barbe (ed.): Very large Scale Integration VLSI. Fundamentals and Applications, Springer Series in Electrophysics, Vol.5 (Springer, Berlin, Heidelberg, New York 1980)Google Scholar
- 1.2F. Rocca: “TV Bandwidth Compression Utilizing Frame-to-Frame Correlation and Movement Compensation”, in Picture Bandwidth Compression, ed. by T.S. Huang, O.J. Tretiak (Gordon and Breach, London 1972)Google Scholar
- 1.6B.K.P. Horn, B.G. Schunck: “Determining Optical Flow; AI Memo 572, M.I.T. (April 1980)Google Scholar
- 1.7D.F. Rogers, J.A. Adams: Mathematical Elements for Computer Graphics (McGraw-Hill, New York 1976)Google Scholar
- 1.8R.Y. Tsai, T.S. Huang: “Three-Dimensional Motion Estimation”, Proc. First European Signal Processing Conference, Sept.16–19, 1980, Lausanne, SwitzerlandGoogle Scholar
- 1.9T. Newman: “Video Target Tracking by Lie Algebra Techniques”, Proc. Workshop on Automatic Missile Tracking, Redstone Arsenal, Alabama, Nov.1979Google Scholar
- 1.10R.J. Schalkoff: “Algorithms for a Real-Time Automatic Video Tracking System”, Ph.D. Thesis, Dept. of Elec. Engr., Univ. of Virginia, Charlottesville, VA (1979)Google Scholar
- 1.11J.K. Cheng, T.S. Huang: “Matching of Relation Structures and Its Application to Image Processing”; Tech. Rpt., School of Elec. Engr., Purdue University (1981)Google Scholar
- J.W. Roach, J.K. Aggarwal: Determining the movement of objects from a sequence of images, IEEE Trans, on Pattern Analysis and Machine Intelligence, Vol.2, No.6, pp.554–562; Nov. 1980Google Scholar
- T.S. Huang, R.Y. Tsai: Three-dimensional motion estimation from image-space shifts, Proc. IEEE International Conf. Acoustics, Speech, and Signal Processing, pp.1136–1139, March 30-April 1, 1981, Atlanta, GeorgiaGoogle Scholar
- R.Y. Tsai, T.S. Huang: Estimating three-dimensional motion parameters of a planar patch, Proc. IEEE Conf. Pattern Recognition and Image Processing, Aug. 3–5, 1981, Dallas, Texas. Amore detailed version to appear in IEEE Trans, on Acoustics, Speech,Google Scholar