Abstract: To overcome the problem that hypersonic targets with different radial distances in near space are difficult to be effectively detected at the same time, this paper proposes a Track-Before-Detect (TBD) algorithm, which improves the accumulation of Hough transform by peak convergence. First, the radial distance-time coordinate is used to describe the data to reduce the influence of measurement errors, and the normalized coordinate allows the two dimensions to be at the same order of magnitude. Hough transform is carried out, and the parameter unit is divided for point accumulation and energy accumulation. The label matrix is used to record the parameter units that each measurement point curve passes through, and all measurement points are traversed, and each measurement point is stored only in its largest parameter unit of the energy accumulation value. New points accumulation and energy accumulation results are then obtained. The threshold is only set for the point accumulation space to extract the target track, and track constraints and merge are performed to output the final track. The simulation results show that the algorithm proposed can detect all three targets with a probability of 90%, demonstrating good multi-target detection performance.

Key words: near space, hypersonic target, peak convergence, Hough transform, track-before-detect, multiple targets