Interference Suppression Algorithm for SAR Based on Time–Frequency Transform

The Rfi Suppression Method Based on STFT Applied to Sar

“…The construction of gain coefficient will exploit the Gaussian statistic characteristic of the SAR echo [8]. The STFT is a linear operation, thus the Gaussian statistic characteristic is reasonable for the time frequency spectrum.…”

“…An adaptive gain coefficient for instantaneous frequency spectrum is derived at every certain time, and it will keep the useful signal correctly during RFI suppression. The approach proposed in this paper has much lower computation load than the approach proposed in [8], and the performance of the proposed approach is demonstrated by both NBI and WBI suppression. This paper is arranged as follows.…”

“…Although the SAR uses 2-dimensional matching filtering, which has inherent ability in interference suppression, the RFI with strong power will still bury useful signal and take on bright stripes along the range direction in SAR images. According to the ratio of interference band to useful signal band, the RFI can be classified into narrow-band interference (NBI) if the ratio is less than 1% and wide-band interference (WBI) if the ratio is greater than 1% [7,8].…”

“…For the instantaneous frequency spectrum estimation of stationary and nonstationary signals, short time Fourier transform (STFT) is an effective tool for time frequency analysis. The RFI suppression method proposed in [8] is the estimation of the instantaneous frequency spectrum by STFT. The strong interference takes on spikes in the instantaneous frequency spectrum, and then the wavelet transform is used to detect the spike and remove it.…”

“…The strong interference takes on spikes in the instantaneous frequency spectrum, and then the wavelet transform is used to detect the spike and remove it. In spite of the good performance of the method proposed in [8] for NBI and WBI suppression, the twice transforms will lead to a very high computation load. Because the interference mostly takes on spikes in instantaneous frequency spectrum, it can be captured directly without wavelet transform, and this will greatly improve the computation efficiency.…”