Range-spread target detection in white Gaussian noise via two-dimensional non-linear shrinkage map and geometric average integration

In this study, a practically efficient method is proposed to detect range-spread targets of manoeuvring flight in white Gaussian noise. High-resolution range profiles (HRRPs) from multiple consecutive pulses are stacked into a two-dimensional (2D) greyscale range-pulse image along range cells and pulses. Based on the local statistics of the images, a 2D non-linear shrinkage map is designed for reducing noise while preserving target echoes before integration. Then, a geometric average (GA) integrator, operating on filtered HRRPs, is designed for target detection. The GA integration exploits the waveform similarity of the filtered target HRRPs. Finally, the experimental results of raw radar data and comparisons with other detectors are reported, showing that the proposed detector can efficiently detect range-spread targets of manoeuvring flight in white Gaussian noise.