Hybrid approach on LIDAR signal processing with information fusion of multiple detectors

In this article, an approach on the LIDAR signal processing to detect low SNR target for the intelligent vehicle is presented. It focuses on the raw data processing of the intensity observations and aims to produce the improved point measurements. The algorithm employs two types of elemental detectors, which are respectively based on the CFAR coherent integration and the Bayesian Track-Before-Detect technique. The raw level information of the detectors before applying the thresholding are delivered to the fusion center, where the joint likelihood ratio hypothesis test is applied to make the final decision on the existence or non-existence of the target. The fundamental experimental results have shown that the proposed fusion approach achieves significant improvement on low SNR target detection and enables to produce further point measurements in comparison to both the conventional detector with constant thresholding and its elemental detectors.