Intrinsic behavioural parameters of autonomous objects based on motion

Abstract

This thesis outlines quantitative behaviour analysis methods. In laboratory conditions, the video tracking-based methods are the best, and thus they are investigated in depth. Methods of image acquisition, physical camera calibration, and spectral image reconstruction are intensively discussed in the author's articles, which are included in the thesis. Software solution for individualized object tracking was developed and applied to aquarium fish. Causality networks were introduced, investigated, and applied to multiple agent data of swimming fish. The networks highlighted school hierarchy, allowed to mine individual parameters of agents, and achieve the primary goal of the thesis -- how individual behaviour manifests itself in collective behaviour.