Spread spectrum photon counting compressed depth imaging method

The pseudo-random spread spectrum photon counting three-dimensional imaging community is able to acquire detailed time-correlated photon counting depth profiles by collecting a large amount of data about a scene and applying the matched filtering (MF) algorithm. However, it limits their ability to process, transmit, and store. To solve this problem, we report the structural time-correlated sparse representation depth reconstruction framework. First, the inversion model to reconstruct the time-correlated photon counting depth profile based on the l1-norm is derived. Second, the Hankel matrix sparsity basis is proposed based on the MF's cross-correlation mechanism. The simulation test results show that compared to the FFT and DCT basis, the proposed basis is sparser. Our main objective is to effectively compress the received photon stamps and exploit the signal sparsity in the correlation domain. In order to validate the proposed framework, extensive experiments on our laboratory system are implemented. The results demonstrate that only 10% of data is sufficient to reconstruct two depth peaks of the partially occluding object with the noise of 100 c/s. When background noise is 100 c/s, 4000 c/s, and 8000 c/s, using the depth estimated by the MF as the reference, the proposed method's imaging mean squared error is 0.3 cm, 1.4 cm, and 2.8 cm, respectively, only with 10%-30% of data. It consumes nearly one ten-thousandth of the energy of the MF method. The proposed framework is excellent in lightweight data process, high speed computation, and low power consumption.