An Adaptive Quantization Method for SAR Target Detection

Due to the limitations of synthetic aperture radar (SAR) beam width and irradiation direction, the radar echo energy varies across different regions in wide scenes, resulting in uneven luminance in SAR images, which affects subsequent SAR target detection and recognition. This paper proposes an adaptive quantization method for SAR images. Firstly, the luminance variation curves are statistically analyzed along the range and azimuth directions to obtain a luminance distribution map. Then, the inverse function of the luminance distribution map is calculated to get a luminance compensation map. Finally, the compensation map is multiplied with the SAR amplitude values to obtain the adaptive quantization results. This paper demonstrates on the real-data SAR-Vehicle dataset that the proposed adaptive quantization method can compensate for the uneven luminance caused by SAR beam drift. The method also improves the target detection and recognition performance of common target detection networks.