Experimentally acquired low-light micro-PIV images of CO2 flows under transcritical and supercritical conditions

This experimentally acquired dataset consists of a series of µPIV-like low-light image sequences of fluorescent particles. The particles were recorded using a high-speed CCD camera and were imaged through a microchannel with flowing carbon dioxide across a range of high-pressure, transcritical conditions inducing optical distortions, so-called speckles. An important feature is that each optically corrupted image sequence is paired with a speckle-free ground-truth of an identical particle pattern. Additionally, the dataset is accompanied by thermodynamic metadata in the form of csv-files for each image.The dataset is designed to support the development, validation, and benchmarking of image reconstruction algorithms under realistic experimental conditions. This can be utilised to drive the advancement of novel data-driven approaches, but also to further develop conventional image reconstruction techniques.Data StructureThe dataset is organised in a hierarchical directory structure designed to provide a clear and traceable mapping between thermodynamic conditions, image data, and associated metadata. At the top level, the data are grouped into individual directories corresponding to sequences. Each sequence represents a distinct experimental protocol (e.g. wall-heater or pre-heater operation) with a unique particle pattern and is listed in the extended operating-point table provided in: 03_additional_data/operating_points_full.csv. Within each sequence directory, sub-directories are organised by operating point. Each operating point corresponds to a unique combination of boundary conditions. Each operating-point directory contains:Ten consecutively recorded image pairs in TIFF formatA comma-separated values (CSV) file containing the associated metadata for each imageThe image naming convention follows the format: SeqXX_OPXX_imgXXX.tif where SeqXX identifies the thermodynamic sequence, OPXX the operating point, and imgXXX the chronological order of acquisition. This filename also serves as the unique image identifier used throughout the dataset. The CSV metadata file links each image identifier to the recorded thermodynamic state variables. All values are reported in SI units. The same folder structure exists for both:raw_images: unprocessed camera outputprocessed_images: images after application of a 99% histogram stretch, as described in the accompanying publicationThe directory 03_additional_data further contains reference image data acquired from fluorescent particles deposited on a simple glass slide. These images can be used to compute particle size distributions and to benchmark the optical and adhesive properties of the particle–adhesive solution independently of the microfluidic experiments.A Jupyter notebook is included with the dataset to demonstrate representative analysis workflows and to support reproducibility. The notebook is located in 03_additional_data/02_code/ and is written entirely in Python.