Fig. 1. Schematic diagram of the photon-level frequency domain dynamic feature SPI. (a) Imaging objects. (b) The reconstructed image with photon counts as observed values. (c) and (d) are 200 and 167 Hz feature images, respectively. (e) Reordered Hadamard cake-cutting pattern. (f) Conventional photon-counting SPI, where the observed values are photon counts y = {y₁, y₂, …, y₄₀₉₆}, corresponding to each pattern. (g) Photon arrival time sequences measured for each pattern. (h) DFT applied on the photon arrival time sequences for each pattern yielded , where each pattern corresponds to a spectrum that includes two featured peaks. The amplitudes of the two peaks in (h) are noted as F₁ = {f_a1, f_a2, …, f_a4096} and F₂ = {f_b1, f_b2, …, f_b4096}. The image reconstruction algorithm is performed using these observed values.

Fig. 2. Schematic diagram of the experimental setup. NDF, neutral density filter; BS, beam splitter; OC, optical chopper; PED, polymer engineered diffuser; SPD, single-photon detector; TDC, time-to-digital converter.

Fig. 3. Photon counting image reconstructions. (a) Physical image of the object captured by a camera and the reconstructed images with different CRs. (b), (c) Relationship of the PSNR and SSIM with the data CR.

Fig. 4. Frequency domain dynamic features SPI results. (a), (b) PSNR and SSIM of images with different CRs at dynamic frequencies of 200 and 167 Hz, respectively.

Fig. 5. Relationship between the dynamic frequency and pattern projection rate of the DMD. (a)–(c) Imaging results of 200 Hz frequency domain imaging at projection rates of 1, 20, and 50 Hz, respectively. (d) Results of the PSNR and SSIM of 200 and 167 Hz frequency domain dynamic feature reconstruction images with the pattern projection rate.