Shadow identification and height estimation of defects by direct processing of grayscale images

Galan, Ulises; Ahuett–Garza, Horacio; Orta, Pedro

doi:10.1007/s00170-018-2933-6

Cited by 5 publications

(2 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thanks to its high speed, low costs, and effectiveness at capturing defects on various surfaces, multi-illumination systems are widely used today in the industry. Applications include detecting casting defects on steel and other metallic surfaces [7,9,11,12,21,26], cracks in paintings [4,16], connection defects in electronic circuit boards [23], or quality control of carbon-fiber materials [18]. Due to the widely varying reflectance properties of the materials involved, defect detection techniques vary substantially in terms of their algorithmic steps.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Defect segmentation for multi-illumination quality control systems

Honzátko

Türetken

Bigdeli

et al. 2021

Machine Vision and Applications

View full text Add to dashboard Cite

Thanks to recent advancements in image processing and deep learning techniques, visual surface inspection in production lines has become an automated process as long as all the defects are visible in a single or a few images. However, it is often necessary to inspect parts under many different illumination conditions to capture all the defects. Training deep networks to perform this task requires large quantities of annotated data, which are rarely available and cumbersome to obtain. To alleviate this problem, we devised an original augmentation approach that, given a small image collection, generates rotated versions of the images while preserving illumination effects, something that random rotations cannot do. We introduce three real multi-illumination datasets, on which we demonstrate the effectiveness of our illumination preserving rotation approach. Training deep neural architectures with our approach delivers a performance increase of up to 51% in terms of AuPRC score over using standard rotations to perform data augmentation.

show abstract

Section: Related Workmentioning

confidence: 99%

“…A second class uses a hand-crafted set of image processing operations to produce feature maps and then use them for defect detection [7,11,18,19]. The operations used include edge-detection, thresholding, as well as various element-wise operations across the images such as maximum, subtraction, or division.…”

Section: Hand-crafted Featuresmentioning

confidence: 99%