Anti-aliasing techniques in photon-counting depth imaging using GHz clock rates
Anti-aliasing techniques in photon-counting depth imaging using GHz clock rates
复制标题
使用 GHz 时钟速率的光子计数深度成像中的抗锯齿技术
DOI:
--
复制
发表时间:
2010
期刊:
影响因子:
--
通讯作者:
G. Buller
中科院分区:
文献类型:
--
作者:
N. Krichel;A. Mccarthy;R. Collins;G. Buller
Single-photon detection technologies in conjunction with low laser illumination powers allow for the eye-safe acquisition of time-of-flight range information on non-cooperative target surfaces. We previously presented a photon-counting depth imaging system designed for the rapid acquisition of three-dimensional target models by steering a single scanning pixel across the field angle of interest. To minimise the per-pixel dwelling times required to obtain sufficient photon statistics for accurate distance resolution, periodic illumination at multi- MHz repetition rates was applied. Modern time-correlated single-photon counting (TCSPC) hardware allowed for depth measurements with sub-mm precision. Resolving the absolute target range with a fast periodic signal is only possible at sufficiently short distances: if the round-trip time towards an object is extended beyond the timespan between two trigger pulses, the return signal cannot be assigned to an unambiguous range value. Whereas constructing a precise depth image based on relative results may still be possible, problems emerge for large or unknown pixel-by-pixel separations or in applications with a wide range of possible scene distances. We introduce a technique to avoid range ambiguity effects in time-of-flight depth imaging systems at high average pulse rates. A long pseudo-random bitstream is used to trigger the illuminating laser. A cyclic, fast-Fourier supported analysis algorithm is used to search for the pattern within return photon events. We demonstrate this approach at base clock rates of up to 2 GHz with varying pattern lengths, allowing for unambiguous distances of several kilometers. Scans at long stand-off distances and of scenes with large pixel-to-pixel range differences are presented. Numerical simulations are performed to investigate the relative merits of the technique.