LSD Based Vision Detection System for Industrial Robot under Complex Illumination Conditions
Weiping Liu, Weidong Chen, Ruimin Wu · 2019
In this paper, we present an object vision detection and localization method approach to the industrial robotic system under complex illumination conditions. In industrial applications of machine vision, many of feature extraction algorithms depend on edge detection, which can be affected by ambient illumination variation and loses efficacy. We propose a robust method that combines image enhancement, edge detection and feature information extraction to detect and locate target components. The key idea is to use a histogram key point constrained homomorphic filtering to enhance the image followed by line and arc segments feature detection. Finally, the experimental results on probe end-face image dataset and steelmaking automation manufacture demonstrate the ability of the proposed method to perform more robust and accurate effectiveness under illumination variant conditions.