Generation of a Segmented Oil Filter Medium from a µCT scan for digital Thickness Estimation
Non-woven materials have traditionally been used in filtration applications, including those for air and oil filters. In this context, we analyze a CT scan of an oil filter medium. The original image has a size of 3828×1376×2509 voxels with a resolution of 750 nm. To ensure accurate thickness estimation, pre-processing is necessary, for which we utilize the GeoDict software.
Initially, the CT scan is imported and rotated so that its primary material axis aligns with the global coordinate system. Following that, the image is cropped to eliminate any empty sections in the plane of the filter. The image is then rotated so that the filter medium lays in the X–Y plane, allowing thickness estimation in the Z direction. To facilitate high-quality segmentation, the non-local means (NLM) filter is utilized to reduce noise in the gray-value image. Subsequently, automatic global thresholding is performed using the Otsu method. The solid material phase of this image consists of cellulose fibers.
Defining the thickness of a non-woven material can be complex due to the protruding fibers and the generally dense region at the top and bottom before reaching the medium's main material density. GeoDict’s Thickness Estimation algorithm (included in MatDict) offers a distinct definition of material thickness, which is vitally important for understanding the filtration efficiency and flow properties of a filter medium. Furthermore, estimating the thickness based on an image can be a valuable tool for quality control of the material.
Citation of article:
Frank, F., Glatt, E., Linden, S., and Wiegmann, A.: Thickness and surface estimation of thin porous media based on 3D image data, 2024, http://dx.doi.org/10.1088/1361-6501/ad2421
The download dataset contains all scan data as well as the result data of the segmentation in GeoDict. The data collection, image processing, and analysis of the metadata are explained in the data description.
Download dataset Data description
The size of the ZIP file is 10 GB. The uncompressed data is about 13 GB.
sha256sum checksum:
2db6f99994d8d0a3724e4e097c835ffdae192f7def5e265d5522b16c7c124b77
Citation of dataset:
Frank F., Glatt E., Linden S., Wiegmann A., 2023: Generation of a segmented oil filter medium from a µCT scan for digital thickness estimation, Math2Market GmbH, Scan data, https://doi.org/10.30423/data.math2market-2023-03.oilfilter.matdict
Citation of article:
Frank, F., Glatt, E., Linden, S., and Wiegmann, A.: Thickness and surface estimation of thin porous media based on 3D image data, 2024, http://dx.doi.org/10.1088/1361-6501/ad2421
