Image Transformation Best Practices
Table of contents
Overview
Image processing is a key element of digitisation. It is a starting point for many ETL processes: High-resolution images have to be acquired without lossy compression (ie. all imaged raw data must be available and no data must be lost because of image compression) at the imaging process to preserve as much information as possible of the specimens.
Recommendations
Image Transformation (TRANSF1, TRANSF2, TRANSF3)
This best practice gives image transformation recommendations.
Level: Basic
Use Case: As a researcher, I want to see specimen images so that I can determine if it can be included to my research.
Recommendation:
TRANSF1: The high-resolution image without lossy compression needs to be acquired in the imaging process to provide a good base for transformation and processing in the later usage.
TRANSF2: Different versions of images, like compressed JPEGs of different spatial resolution and small scale 3D models from CT scanning data, need to be extracted for different purposes such as OCR, online viewing and sharing.
TRANSF3: For long-term data preservation, usually TIFF images are used and proprietary image formats like Nikon’s NEF and Canon’s CR2 raw images may not be supported.
Discussion
For conventional 2D imaging, high-resolution TIFF or RAW images have large file sizes up to hundreds of MB. TIFF images with lossless compression can reduce half of the uncompressed TIFF file size. Converting 12/14/16-bit images to 8-bit will reduce the file size but will also lose colour and tone information. The same image in the JPEG format is usually less than 10 MB file size. However, the lossy compression used in JPEG results in the loss of information in the image. And reducing the spatial resolution of the image will further reduce the file size, such as thumbnail images with only tens of KB file size.
Imaging process is one of the key parts in the digitisation process. It not only provides images for viewing but also is a starting point for many ETL processes. High-resolution images have to be acquired without lossy compression (ie. all imaged raw data must be available and no data must be lost because of image compression) at the imaging process to preserve as much information as possible of the specimens. This will provide a good base for various later use cases. Usually TIFF images or RAW images like Nikon’s NEF and Canon’s CR2 formats are used.
3D imaging acquires large amounts of raw data from the imaging devices. For microCT scanners, data is up to hundreds of GB. Data transformation and processing have to be done to generate a 3D model with tens of GB file size. To further reduce the file size, a small scale version can be extracted from the 3D model.
Therefore, depending on the use cases, the original acquired images will be transformed into different versions with different image formats, spatial and colour resolutions. Original images will be needed in tasks that require accurate information of the objects in the image, such as quality check, OCR, and segmentation in the digitisation process. JPEG images with the full spatial resolution can be used for online sharing. For online viewing, images have to be transformed into JPEG images with different spatial resolutions, like the small thumbnail images. Similarly, small scale 3D models are needed for online viewing.
For long-term data preservation, original data has to be kept. Usually TIFF images are used and proprietary image formats like Nikon’s NEF and Canon’s CR2 raw images may not be supported.
Implementation Example:
Finnish Museum of Natural History (Luomus)
High-resolution TIFF images without compression are acquired from mass digitisation systems. At the imaging stations, JPEG images with the same spatial resolution are converted from TIFF images and the TIFF images are lossless compressed. Different spatial sizes of JPEG images are generated for online viewing and sharing. All images are achieved for backup and will be stored in the planned long-term preservation.
Authors
Zhengzhe Wu and Esko Piirainen
Finnish Museum of Natural History (Luomus)
Contributors
Lisa French, Laurence Livermore
Citation
References
Licence
Document Control
Version:
Changes since last version: N/A
Last Updated: 22 June 2022
Edit This Page
You can suggest changes to this page on our GitHub