On parameterizing higher-order motion for behaviour recognition

Human behaviours consist different types of motion; we show how they can be disambiguated into their components in a richer way than that currently possible. Studies on optical flow have concentrated motion alone without the higher order components: snap, jerk and acceleration. We are first to acceleration, jerk, snap constituent parts obtained from image sequences, deployed for analysis, especially behaviour. demonstrate estimation acceleration sport, human motion, traffic scenes violent behaviour wide potential application analysis Determining is suited which contain them: innate containing acts behaviour, but it not just contains quickly changing movement: gait though approaches yet consider radial tangential since concentrate alone. The synthetic real-world images illustrates ability discriminate objects under motion. Then new applied heel strike detection gait. These results approach ready developing applications recognition provides basis future research higher-order analysis.