A novel approach on classification of infant activity post surgery based on motion vector

摘要

Pain is a subjective experience and no objective test exist to measure it, IASP (International Association for the Study of Pain) decided patients self-report as gold standard of pain assessment. Infant cannot provide a self-report of pain verbally. In this paper, we developed a system to recognize post-surgery infant activity based FLACC (Face, Legs, Activity, Cry, Consolability), score 0 is given if the infant moves easily, score 1 if the infant is squirming, and a score 2 if the baby is jerking by observing the features of motion. In FLACC, activity is one of the five parameters to identify the level of infant pain. Using a block matching algorithm with the addition of the Taylor series to generate a value with sub-pixel motion accuracy from the reference frame to the current frame in the form of a motion vector. Videos have been verified by doctors and nurses using hormone cortisol with FLACC measurements. The results of the experiment show the classification using SVM could detect the infant activity moves easily, squirming, and jerking at 90.4762%. Nevertheless, this experiment is still novel and needs further study on the infant activity.