Photogrammetry and recognition of gait patterns are valuable
tools to help identify perpetrators based on surveillance
recordings.
We have found that stature but only few other measures have
a satisfying reproducibility for use in forensics.
Several gait variables with high recognition rates were
found. Especially the variables located in the frontal plane
are interesting due to large inter-individual differences in
time course patterns.
The variables with high recognition rates seem preferable
for use in forensic gait analysis and as input variables to
waveform analysis techniques such as principal component
analysis resulting in marginal scores, which are difficult to
interpret individually.
Finally, a new gait model is presented based on functional
principal component analysis with potentials for detecting
individual gait patterns where time course patterns can be
marginally interpreted directly in terms of the input variables.
In this presentation, the above methods will be discussed
exemplified with forensic cases.