Sub-sequence principal component approximation denoising for radar high resolution range profiling

Radar high resolution range profiling is an effective means for target feature analysis and target recognition. In practice, radar return is usually contaminated by strong noise and clutter. This results in profile distortion and recognition performance degradation. To improve the quality of profile, in this paper, we present a new denoising method to improve the signal-to-noise ratio (SNR) of the return. The return is segmented into sub-sequences by sliding-window to construct sub-sequence matrix. We demonstrate that the signal components and noise part can be separated in subspace. The SNR improved return is recovered by the sub-sequence principal components approximate. Experimental results show that this approach can effectively enhance the SNR, leading to a high-quality profile.