An analysis of feature detection efficiency on images with a priori unknown and changing viewing conditions

摘要

A comparative analysis of 3×3 pixel robust detector and Harris detector in terms of stability and performance detection in case of viewing conditions change caused by the collection and conversion of image cameras presented. The efficiency of detectors was evaluated in terms of the repeatability of feature detection on a different pair of images when the conditions of observation, camera parameters (focal length, aperture size of the lens), camera position relative to the observed scene, scene illumination changes. Simulation results show that a robust detector used under these conditions provides a greater number of stable features of the image detected.