LOG ENVELOPE LIKELIHOOD RATIO BASED VELOCITY FILTERING TRACK-BEFORE-DETECT METHOD

Track-before-detect (TBD) method has an advantage in detecting and tracking weak targets, due to the direct use of raw sensor data. Velocity filtering based TBD (VF-TBD) is a typical multiframe TBD method, along with the benefit of high energy integration efficiency and well-focused output envelope. However, in conventional VF-TBD methods, the statistical information of raw data is not used for multiframe integration, which may cause performance degradation. In this paper, log envelope likelihood ratio based VF-TBD is proposed, in which the statistical information of raw data is taken into account adequately for multiframe integration. To deal with the complex and real valued data efficiently, the logarithm of complex envelope likelihood ratio (LCELR) and the logarithm of real envelope likelihood ratio (LRELR) are utilized in multiframe integration of VF-TBD, respectively. The calculations of LCELR and LRELR are presented, and the procedure for multiframe processing of the proposed VF-TBD is provided in detail. Numerical experiments are conducted to compare the proposed methods with conventional TBD methods in terms of computational efficiency, integrated envelope, detection performance and estimation accuracy.