Cargando…

Surface Defects Detection of Stamping and Grinding Flat Parts Based on Machine Vision

Currently, surface defect detection of stamping grinding flat parts is mainly undertaken through observation by the naked eye. In order to improve the automatic degree of surface defects detection in stamping grinding flat parts, a real-time detection system based on machine vision is designed. Unde...

Descripción completa

Detalles Bibliográficos
Autores principales: Tian, Hongzhi, Wang, Dongxing, Lin, Jiangang, Chen, Qilin, Liu, Zhaocai
Formato: Online Artículo Texto
Lenguaje:English
Publicado: MDPI 2020
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7472636/
https://www.ncbi.nlm.nih.gov/pubmed/32823558
http://dx.doi.org/10.3390/s20164531
Descripción
Sumario:Currently, surface defect detection of stamping grinding flat parts is mainly undertaken through observation by the naked eye. In order to improve the automatic degree of surface defects detection in stamping grinding flat parts, a real-time detection system based on machine vision is designed. Under plane illumination mode, the whole region of the parts is clear and the outline is obvious, but the tiny defects are difficult to find; Under multi-angle illumination mode, the tiny defects of the parts can be highlighted. In view of the above situation, a lighting method combining plane illumination mode with multi-angle illumination mode is designed, and five kinds of defects are automatically detected by different detection methods. Firstly, the parts are located and segmented according to the plane light source image, and the defects are detected according to the gray anomaly. Secondly, according to the surface of the parts reflective characteristics, the influence of the reflection on the image is minimized by adjusting the exposure time of the camera, and the position and direction of the edge line of the gray anomaly region of the multi-angle light source image are used to determine whether the anomaly region is a defect. The experimental results demonstrate that the system has a high detection success rate, which can meet the real-time detection rEquation uirements of a factory.