Comparison of genetic algorithm components and selection variants in unlawful behavior detection of hand movement

Abstract

Video Surveillance System (VSS) relies on behavior detection mostly for abnormal or unlawful gestures. Nowadays, the manual detection of unlawful or suspicious behavior through the ongoing advancement of the size of the image that need to be scan through. This paper aims to present an unlawful behavior detection of hand movement which apply genetic algorithm (GA) to efficiently determine the validity of the hand movements. An investigation on GA components was performed to determine which component has the strongest impact on the performance. GA selection was identified; tournament selection and random permutation selection were chosen to be experimented. The result demonstrates tournament selection produced better result in terms of the detection rate and false negative rate by using confusion matrix derivation. Video Surveillance System (VSS) relies on behavior detection mostly for abnormal or unlawful gestures. Nowadays, the manual detection of unlawful or suspicious behavior through the ongoing advancement of surveillance cameras make it become unfeasible due to the size of the image that need to be scan through. This paper aims to present an unlawful behavior detection of hand movement which apply genetic algorithm (GA) to efficiently the hand movements. An investigation on GA components was performed to determine which component has the strongest impact on the performance. GA selection was identified; tournament selection and random permutation selection were chosen . The result demonstrates tournament selection produced better result in terms of the detection rate and false negative rate by using confusion matrix derivation. unlawful behaviour; genetic algorithm; tournament selection; random