This repository combines methods for object detection and direct image feature extraction for crystallisation image datasets using MATLAB. Currently, the codebase is tailored to routinely process image and sensor data from Crystalline (Technobis Crystallization Systems, Alkmaar, The Netherlands), however, functions for object detection and direct image feature extraction can be flexibly applied to other image datasets.
- _Demo: contains specific demo files and demo image data (.zip).
- _FileExchange: dependencies, please review individual file packages (Readme). Some files were modified to improve integration.
- CryExp_Masterfile.xlsx: contains information with metadata, data paths and image processing parameters.
- Crystalline_Main.m: main MATLAB file to run Crystalline image analysis routine.
- Crystalline_ImgAnlys.m: image analysis routine which includes Object Detection (Crystalline_ImgPrc), Direct Image Feature Extraction (Crystalline_ImgAnlys_Feat) and Object Parameterisation (Crystalline_ImgAnlys_ObjPara).
- Crystalline_DataAnlys.m: data analysis for post-processing of extracted direct image features for clear/cloud point detection and plotting.
To add your own datasets to the image analysis:
- Open parameter file 'CryExp_Masterfile.xlsx' and amend/add row with details on your image dataset: see sheet 'Crystalline' for details on image dataset, see 'Prc_ImgAnlys' for image processing parameters
- Open the main file (Crystalline_Main.m) and run the script. The datasets will be analysed according to their order in 'CryExp_Masterfile.xlsx\Crystalline'. The MATLAB analysis will run through the datasets top to bottom until it encounters an empty row.
- Frederik J. S. Doerr, Cameron J. Brown, and Alastair J. Florence (2022). Direct Image Feature Extraction and Multivariate Analysis for Crystallization Process Characterization. Cryst. Growth Des. https://doi.org/10.1021/acs.cgd.1c01118
Data repository: https://doi.org/10.15129/75002e1c-80dc-42ab-984a-3444fbef9c03