High resolution single pixel imaging

This dataset contains the data used to produce the results from the manuscript:Stojek, R.; Pastuszczak, A.; Wróbel, P.; Cwojdzińska, M.; Sobczak, K.; Kotyński, R.: "High-Resolution Single-Pixel Imaging of Spatially Sparse Objects: Real-Time Imaging in the Near-Infrared and Visible Wavelength Ranges Enhanced with Iterative Processing or Deep Learning". Preprints2024, 2024111153. https://doi.org/10.20944/preprints202411.1153.v1The data includes image maps, the measurement matrix with binary sensing patterns and the generalized inverse of the measurement matrix for single pixel imaging at the native resolution of the DMD, i.e. 1024 x 768 pixels.It also includes the trained model of the neural network, which has been used in the manuscript to enhance the image reconstruction, with a portion of a Python code which enables loading all the necessary data, performing simulated single-pixel measurement and the initial reconstruction with the provided measurement matrix and its inverse, and finally performing the 2-nd stage reconstruction with the neural network and visualizing the results. We also include 2 sets of handcrafted testing images used in the above manuscript, each containing 5000 images with resolution 1024 x 768 of either high-resolution binary line-art or modified MNIST dataset.