Influence of some acquisition parameters in non-line-of-sight imaging

Over the past few years, Non-Line-of-Sight (NLOS) scene reconstruction algorithms have made significant progress, transforming our ability to reconstruct environments and scenes that are hidden from the direct view. These algorithms employ sophisticated techniques such as f-k migration, back-propagation, or phasor-field models to reconstruct the scene and reveal hidden objects. However, a major challenge persists: the time required to acquire the vast amounts of data needed to generate a high-quality, high-resolution scene reconstruction. High-resolution scene reconstructions necessitate extensive data collection, often involving measurements of multi-scattered light, leading to substantial data volumes. This results in prolonged acquisition times and considerable computational costs. Therefore, a pressing issue is how to reduce acquisition time without sacrificing the accuracy and resolution of the final reconstruction. To tackle this challenge, we have systematically analyzed the impact of each of these acquisition parameters to understand their influence on both acquisition time and data-transient quality. Our goal is to identify the most significant parameters and adjust them to achieve the desired trade-off between speed and accuracy. One of the most pressing issues is how to reduce the acquisition time without compromising the accuracy and resolution of the final reconstruction. To get closer to real-time processing, it is essential to carefully consider the parameters involved in the acquisition phase. Key parameters including the sensor pixel calibration, the exposure time and number of images per frame, the camera field-of-view (FOV) and focus adjustment, all play a critical role in both the quality of the reconstruction and the time required to acquire the necessary data. Optimizing these acquisition parameters is key to improving the efficiency of the scene reconstruction process.