COMPARISON OF THE CONTOUR INFLECTION POINTS SELECTION METHODS FOR VIDEO SURVEILLANCE SYSTEMS

Abstract

Comparison of the contour inflection point selection of the: interpolation and differential methods and the method based on wavelet analysis of the function curvature were performed in the article according to the following criteria: probability of the correct detection, probability of the false detection and an error in coordinate evaluation for systems of video vision that are based on the model task results. Work results of the methods were analyzed and graphically presented. It was demonstrated that for simple geometrical figures the interpolation method has low accuracy (it displaces inflection points). And differential method in its turn possesses the best accuracy indices though it selects excessive inflection points (it has low obstacle resistance). And the method based on wavelet analysis of the function curvature shows best results in noisy conditions.

Keywords

• системи відеоспостереження, контур зображення, характерні точки, методи виділення характерних точок (диференціальний, інтерполяційний, вейвлет-аналіз функції кривизни)