A Structure-Aware Compressed Sensing Framework for Robust Multipath Suppression in Indirect ToF Imaging

We propose a novel compressed sensing method to improve the depth reconstruction accuracy and multi-target separation capability of indirect Time-of-Flight (iToF) systems. The method constructs a sparse observation model using a single modulation frequency with multiple phase shifts and narrow-duty-cycle continuous waves, avoiding the complexity of multi-frequency schemes and data-driven reconstruction while preserving strong measurement sensitivity. During sensing matrix construction, pixel-wise path-length variations induced by lens distortion are incorporated to better align the model with the actual modulation response. To enhance sparse recovery, we first cluster the distance-response dictionary using K-means. Then, during the Orthogonal Matching Pursuit (OMP) step, atom selection is restricted within each cluster, which reduces the search space and improves reconstruction stability. Experimental results show that the proposed approach reduces depth reconstruction error to the millimeter scale and significantly improves multi-target separation in typical multipath scenarios, achieving superior accuracy and robustness compared with existing methods. The entire framework requires no hardware modification and can be seamlessly integrated into current iToF systems.