Time-resolved data analysis framework for ImageJ: Motion, Tracking, Segmentation

Abstract:

Motion estimation is an important field in Computer Vision. Optical flow methods determine a displacement ﬁeld which maps all image points from the ﬁrst frame to their new locations within the second frame. This field can be used for a variety of tasks such as velocimetry and temporal data analysis.

This work aims to combine a number of methods based on motion estimation. These methods include: motion analysis and visualization, an automated or semi-automated tracking, motion-based segmentation and capture of temporally changed data. For all of these techniques a motion estimation method can serve as a core which unifies presented techniques into a single framework for time-resolved data analysis.

The proposed motion estimation technique is based on variational optic flow methods with a coarse-to-fine strategy employed. The method for optical flow estimation is flexible and can be adopted to different image features. It could be used for any type of motion as well as for large displacements within an image domain. The resulting displacement field is dense and allows subpixel precision.

Keywords:
Plugin, time-resolved data, motion estimation, motion analysis, optical flow, variational methods, tracking, motion-based segmentation

Author
Alexey Ershov

Organisation
ISS, Forschungszentrum Karlsruhe GmbH, Germany

Homepage

Short Biography

Master of Computer Science

University of Saarland, Saarbrucken, Germany

Software Development Engineer

Khabarovsk State University of Technology, Khabarovsk, Russia

Research interests:

Image Processing and Computer Vision
Algorithms and Data strucures
2D \ 3D radiographic image analysis
Motion estimation and analysis
Time-resolved data analysis