Unique Parametric Ratio Measure for Geometrical Shape Object Detection

摘要

The decision-making methodology of an object classifier is based on the values of the measured parameters. The paper proposes an algorithm for target classification with video-based applications, based on ratios of parameters extracted using blob detection. The technique has been developed for different kinds of parametric ratios like ratio of area/major axis, area/minor axis, major axis/minor axis. Whereas, a unique parametric ratio i.e. the ratio of area/minor axis has a specific significance for classifying different categories of objects. The algorithm will differentiate among four major classes of objects i.e. human, two-wheelers, four-wheelers and heavy vehicles like Trucks. The algorithm is trained and tested on videos recorded by day cameras and thermal cameras. The experimental results are shown in tabular form, which are also validated on the basis of published work.