Multireturn compressed gated range imaging

Abstract

Abstract. Active range imaging (RI) systems utilize actively controlled light sources emitting laser pulses that are subsequently recorded by an imaging system and used for depth profile estimation. Classical RI systems are limited by their need for a large number of frames required to obtain high resolution depth information. In this work, we propose an RI approach motivated by the recently proposed compressed sensing framework to dramatically reduce the number of necessary frames. Compressed gated range sensing employs a random gating mechanism along with state-of-the-art reconstruction algorithms for the estimation of the timing of the reflected pulses and the inference of distances. In addition to efficiency, the proposed scheme is also able to identify multiple reflected pulses that can be introduced by semi-transparent elements in the scene such as clouds, smoke, and foliage. Simulations under highly realistic conditions demonstrate that the proposed architecture is capable of accurately recovering the depth profile of a scene from as few as 10 frames at 100 depth bins resolution, even under very challenging conditions. The results further indicate that the proposed architecture is able to extract multiple reflected pulses with a minimal increase in the number of frames, in situations where state-of-the-art methods fail to accurately estimate the correct depth signals. IntroductionRange imaging (RI) refers to a family of technologies that aim to capture and extract the depth information of a scene. Formally, assuming that each point in a scene imaged by the sensor is characterized by a single depth value (the distance between the object and the camera), one can generate a depth map of a scene where pixel intensity corresponds to the distance between the camera and the imaged object. Depth information can be directly used for the visualization of the scene, as well as for target detection, robot navigation, and surface modeling. Today, several RI solutions are available in the market depending on the hardware requirements, the physical constraints, and the requested depth data quality and resolution. RI has been employed in numerous applications including remote sensing, gaming, security, search and rescue, and medical diagnostics.

Sparsity-Aware 3-D ToF Sensing

“…He obtained some remarkable results through numerical simulations by using OMP and evaluating two different resolutions, 128 and 256 depth bins, for several sampling rates and noise levels. The aforementioned study was extended to the retrieval of 2-sparse signals in 2015 [49], by using non-negative LASSO [17] and OMP. They obtained exact retrieval from a sampling rate of r = 10%, by exploiting the prior knowledge of the shape of the cross-correlation function.…”

Sparsity-aware 3D ToF Sensing

“…He obtained some remarkable results through numerical simulations by using OMP with 10 iterations per reconstruction and evaluating two different resolutions, 128 and 256 depth bins, for several sampling rates between 5% and 49% and different noise levels. The aforementioned study was extended in 2015 [15], by using non-negative least absolute shrinkage and selection operator (LASSO) [16] and OMP [17], to the retrieval of 2-sparse signals which considered the presence of an additional translucent object in the scene, obtaining exact retrieval from a sampling rate of r = 10%, by exploiting the prior knowledge of the shape of the cross-correlation function. Another noticeable work in the field of coded ToF cameras was the one performed by Kadambi in 2013 [18] by undertaking sparse deconvolution to reconstruct sequences of Diracs corresponging to multipath returns relying on pulseshaped autocorrelation function.…”

Recent advances in transient imaging: A computer graphics and vision perspective

“…In order to improve convergence on range imaging (i.e. focusing on single scattering), approaches such as rangegates coding were developed, where the gate response is continuously modulated over a long time window, then reconstructed using intensity analysis [Laurenzis et al 2007;Zhang and Yan 2011;Laurenzis and Bacher 2011], compressed sensing (CS) on the temporal domain by random temporal gating [Li et al 2012;Tsagkatakis et al 2012;Tsagkatakis et al 2013;Tsagkatakis et al 2015], or hybrid approaches combining both techniques [Zhang et al 2012;Dai et al 2013]. This can reduce the number of measurements to just two, for 13-bit range images.…”