Subject-object discrimination in 4D dynamic scene interpretation for machine vision

Abstract

A concept for subject-object discrimination has been developed, based on the integral spatiotemporal world model for machine vision. The control determination by feedback of estimated state variables leads directly to behavioral competences for classes of situations. Subjects are defined to have controls at their disposal, which allow them to freely initiate motion activities without depending on changes in the environment. The proposed concept pertains to the classification of physical objects in the limited domain of road traffic. Reference is made to experimental data on VAMORS (vehicle for autonomous mobility through computer vision).<>