Analysis of Three Dimensional Recovery Algorithms' Influence on the Ranging Accuracy of Gm-APD Ladar

Shang Gao,Jianfeng Sun,Peng Jiang,Di Liu,Xin Zhou,Hongming Fu,Penghui Wang
DOI: https://doi.org/10.1117/12.2504677
2018-01-01
Abstract:Geiger-mode Avalanche Photo Diode(Gm-APD) Ladar is a probabilistic device outputting three dimensional(3D) images based on the multi-frame imaging statistics, which makes the 3D recovery algorithm one of the key techniques of imaging system. Besides, performance of algorithm plays a crucial role in improving recovery quality of 3D images. This paper researches three 3D algorithms based on histograms, containing peak-selecting algorithm, range-selecting algorithm and Gaussian fitting algorithm. Firstly, the triggered model of Gm-APD is analyzed based on the work timing sequence and imaging theory of Gm-APD Ladar. Meanwhile, the recovery principles of three algorithms are analyzed and clarified. Secondly, two evaluation criterions, average range error and accuracy of range recovery, are raised to evaluate range accuracy. Finally, range images are obtained with above three algorithms in statistics of different frames, based on original data obtained from 64 x 64 Gm-APD Ladar imaging experiment. With the three construction algorithms, the result shows that the range accuracy of recovery range images improves and converges to 0.2 similar to 0.3m with the increment of number of frames participating in the statistics, and the accuracy of range recovery can be up to 90%. In low frame numbers, the range accuracy of recovery range profile is worst with peak-selecting algorithm, and the average range error with range-selecting algorithm performs best while accuracy of range recovery with Gaussian fitting algorithm is highest among all algorithms. The result has important guiding significance for the choice of recovery algorithm under different requests.
What problem does this paper attempt to address?