Non-line-of-sight (NLOS) imaging reconstructs scenes hidden from direct view by analysing light that has scattered off visible surfaces. In practice, those measurements carry noise and deviate from the idealized light transport that reconstruction algorithms typically assume. According to the paper, these effects can degrade reconstructions produced by conventional light-cone transform (LCT) methods.
To address this, the authors propose a measurement-adaptive light-cone transform. The title and abstract indicate that the transform is adjusted to the measured data, rather than relying on a fixed idealised model, with the goal of making reconstruction robust under realistic conditions. The available abstract does not specify the exact mechanism.
Because this is a single source, there is no independent result to compare against. The contribution, as stated, is a robustness improvement for NLOS imaging in settings where measurements are imperfect.