Moving Object Trajectory Mining Moving Object Trajectory Mining - PowerPoint PPT Presentation

Moving Object Trajectory Mining Moving Object Trajectory Mining Trajectory decomposition algorithm Trajectory decomposition algorithm Somayeh Dodge Robert Weibel Representation Analysis and Representation, Analysis and Visualization of Moving Objects Dagstuhl, November 2008

Initial idea � How similar proxy MPOs are to the corresponding reference MPOs di f MPO � HCI studies (mouse movement as a proxy of ( p y eye movement) (Chen et al,2001; Cox et al, 2006) � Social science studies (bank note dispersal as � Social science studies (bank note dispersal as human movement) (Gonzàles et al,2008) � Main goal: to reveal similarities between � Main goal: to reveal similarities between physics of movement of different objects

Trajectory decomposition j y p � Aim: Local movement feature extraction � Extracting movement parameters from trajectories of � E t ti t t f t j t i f MPOs � Input: � Movement parameter profile (e.g. speed, acceleration, p p ( g p , , straightness index, turning angle) � Output: � Output: � Segments of homogeneous movement characteristics

Decomposition algorithm p g � Key elements: � Key elements: � Variation in sinuosity � Variation in deviation from median line � Variation in deviation from median line

Decomposition outcome p Dodge, S., Weibel, R., Forootan, E., (2008): Revealing the Physics of Movement: Comparing the Similarity of Movement Characteristics of Different Types of Moving Objects.. Under review of the "Spatial Data Mining: Methods and Applications", A Special Issue of the Journal of Computers, Environment and Urban Systems.

Applications (1) pp ( ) � Classifying trajectory data generated by unknown moving objects and assigning k i bj t d i i them to known types of moving objects. yp g j � Similarity analysis tasks � Finding similarities in movement behavior of � Finding similarities in movement behavior of moving objects from the same or different types

Applications (1) - example pp ( ) p Dodge, S., Weibel, R., Forootan, E., (2008): Revealing the Physics of Movement: Comparing the Similarity of Movement Characteristics of Different Types of Moving Objects.. Under review of the "Spatial Data Mining: Methods and Applications", A Special Issue of the Journal of Computers, Environment and Urban Systems.

Applications (2) pp ( ) � Trajectory simulation � � ... in ecological modeling (Blythe et al., 1996; Turchin, in ecological modeling (Bl th t l 1996 T hi 1998) � ... in spatio-temporal database research (Pfoser & � in spatio temporal database research (Pf & Theodoridis, 2003) � � ... in data mining algorithm evaluation (Laube & in data mining algorithm evaluation (Laube & Purves, 2006) � Reproducing MPOs according to the global and local characteristics of objects having the same local characteristics of objects having the same dynamic behavior

Applications (3): Saccade and fixation d t detection on eye movement protocols ti t t l Fi Fixation ti Saccade

Thanks for your attention! y