A Novel Approach to Real-Time Motion Detection

Abstract

A real-time scheme is introduced that meets a need for fast, reliable, multiple-target motion detection without accurate object velocity or structure measurements. A multiresolution approach decomposes images into a pyramid with several spatial frequency bands to selectively detect motion of interest. Multiple moving targets are detected using a multiple window, coarse-to-fine focus of attention scheme to reconstruct motion energy and search for targets. Real-time sensor motion (translation and rotation) compensations use hierarchical correlation and minimum perturbation. Results are shown for multiple moving targets in FLIR (forward-looking infrared) image sequences from both stationary and moving sensors.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Cite

Text

Lee and Lin. "A Novel Approach to Real-Time Motion Detection." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1988. doi:10.1109/CVPR.1988.196315

Markdown

[Lee and Lin. "A Novel Approach to Real-Time Motion Detection." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1988.](https://mlanthology.org/cvpr/1988/lee1988cvpr-novel/) doi:10.1109/CVPR.1988.196315

BibTeX

@inproceedings{lee1988cvpr-novel,
  title     = {{A Novel Approach to Real-Time Motion Detection}},
  author    = {Lee, James S. J. and Lin, Charlotte},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {1988},
  pages     = {730-735},
  doi       = {10.1109/CVPR.1988.196315},
  url       = {https://mlanthology.org/cvpr/1988/lee1988cvpr-novel/}
}