Near-Infrared, Depth, Material: Towards a Trimodal Time-of-Flight Camera

Time-of-Flight cameras are active sensors that able to capture both the light intensity reflected by each observed point in scene and distance between these points camera. Enhancing images with a depth modality enables capturing surfaces 3D boosts applicability of sensors. Nevertheless, high-level information still needs be extracted from data stream order accomplish tasks, like recognition or classification. Ideally, semantic gap sensor output requirements should as small possible, reduce computational cost failure probability. An additional helps this regard, but there further cues can seen ToF have remained underexploited so far. In paper we take first steps towards trimodal camera, which adds valuable xmlns:xlink="http://www.w3.org/1999/xlink">material modality classical modalities. To end, raw is used obtain Fourier samples impulse response function (MIRF) modulated illumination. The MIRF depends on surface- sub-surface-level scattering mechanisms material and, thus, identify materials different nature. Consequently, distinctive feature vectors obtained measurements. Additionally, including incidence angle allows dependency parameter. Experimental validation confirms feasibility approach. We also constructed live demonstrator our