High-frame-rate and large-field-of-view relay-projection microscopic telescopy via resampling

Microscopic video imaging techniques that combine high frame rates with large fields of view (FOV) are essential in applications ranging from biomedical microscopy to industrial inspection. However, simultaneously achieving both high frame rates and large FOVs presents significant challenges. One strategy involves on-chip pixel binning—a process essentially equivalent to downsampling—that boosts the frame rate of area detectors while improving the signal-to-noise ratio in optical imaging. In the context of relay-projection microscopic telescopy (rPMT), a novel computational microscopic imaging technique, pixel binning increases the frame rate but at the cost of enlarging the sampling interval in the Fourier spectrum plane. Since the object plane and the Fourier spectrum plane are linked by Fourier transform, the FOV in the object plane is consequently reduced, potentially failing to cover the entire object, leading to information loss and possible object reconstruction failure. A similar issue may arise when using detectors with inherently large pixel sizes. To address the issue of reduced FOV in the object plane, we propose applying interpolation-based upsampling to the spatial power spectrum images during preprocessing. This adjustment of the sampling interval in the Fourier spectrum plane effectively expands the object plane's FOV, optimally utilizes the spatial power spectrum information, and facilitates successful object reconstruction. We first validated the feasibility and effectiveness of this method experimentally. Subsequently, we analyzed the impact of the upsampling rate on reconstruction quality and developed a procedure for estimating the optimal upsampling rate to achieve high-efficiency, high-quality reconstructions. Finally, we conducted a comparative analysis of various interpolation algorithms, such as nearest, bilinear, and bicubic interpolation, and their effects on object reconstruction outcomes. By integrating on-chip pixel binning with interpolation-based upsampling, imaging frame rates of rPMT can be enhanced while maintaining a consistent FOV in the object plane. This technique holds significant potential for advancing dynamic microscopic video imaging methods with high frame rate and large FOV, thereby contributing to the evolution of computational microscopic imaging technologies.