The Analysis Of Multiresolution Employed In UWB SAR Target Detection

Ultra-wide band synthetic aperture radar (UWB SAR) foliage-concealed target detection technology is the basis of UWB SAR ATR and the important part of the UWB SAR system. The detection performance in the condition of multiresolution is usually better than the performance in the condition of single-resolution. Hence, this paper tries to enhance the performance of UWB SAR target detection through the use of multiresolution analysis.The multiresolution statistic feature extraction used in UWB SAR target detection is investigated. The multiresolution characteristics of the trunk clutter and target are analyzed, which is the foundation of the multiresolution statistic feature extraction. The applicability of multiresolution statistic feature extraction methods suitable in high-frequency SAR image are analyzed when they are used in UWB SAR image. With the analysis of the characteristic of the trunk clutter and target in the low-frequency UWB SAR, a multiresolution statistic features extraction approach that suited for low-frequency UWB SAR target detection is proposed.The multiresolution energy feature extraction in UWB SAR target detection is investigated. Based on the UWB SAR echo model and image principle, the frequency domain of UWB SAR image is deduced. A new multiresolution imagery decomposition algorithm is presented with the analysis of the characteristic of the trunk clutter and target. The corresponding filters are offered. Then multiresolution energy feature that realizes target detection is extracted through multiresolution imagery serial.The multiresolution CFAR detection based on wavelet transform in UWB SAR target detection is investigated. The feasibility of the wavelet transform used in UWB SAR image target detection is analyzed. A auto-adapted MK-S test method is presented for the recognition of UWB SAR clutter. Then a CFAR detection approach is proposed based on multiresolution image.Finally, the research prospect in future is presented based on the study summary of full paper.