Hadamard Matrix Slicing Single‐Pixel Imaging : Establishing a Linear‐Array‐Inspired Paradigm for N ‐fold Acceleration Single‐Pixel Imaging

Li, Xiaoxue and Hou, Jieting and Zhao, Qizheng and Jia, Yiqi and Zhong, Xiang and Ma, Mengchao and Deng, Huaxia and Gong, Xinglong and Wang, Ziwei (2026) Hadamard Matrix Slicing Single‐Pixel Imaging : Establishing a Linear‐Array‐Inspired Paradigm for N ‐fold Acceleration Single‐Pixel Imaging. Advanced Science: e75643. ISSN 2198-3844

Text (advs.75643.pdf)
advs.75643.pdf - Published Version
Available under License Creative Commons Attribution.
Download (7MB)

Abstract

Linear array detectors dominate fields like industrial inspection owing to their efficient push‐broom scanning modes. However, Single‐Pixel Imaging (SPI) remains constrained by global 2D modulation, where the requisite pattern count scales quadratically with resolution. This imposes severe bottlenecks on data throughput and modulation duration. To overcome these limitations, Hadamard Matrix Slicing Single‐Pixel Imaging (HMS‐SPI) is proposed, which establishes an efficient 1D imaging paradigm analogous to linear array sensors. By mathematically slicing the traditional 2D Hadamard matrix into 1D encoding vectors, HMS‐SPI reduces the measurement patterns required for an N × N $N \times N$ image by a factor of N $N$ while maintaining full sampling. Experimentally, a 768 × 512 $768 \times 512$ image is reconstructed from 1,024 measurements in 12.8 seconds at 80 Hz. Furthermore, theoretical analysis demonstrates a potential 51.2 milliseconds acquisition time at a 20 kHz modulation limit. This approach drastically reduces memory consumption and pattern loading latency. By emulating the efficiency of push‐broom scanning, HMS‐SPI provides a transformative solution for high‐throughput SPI applications.