论文部分内容阅读
为了实现星载合成孔径雷达(synthetic apertureradar,SAR)成像处理器小型化和实时信号处理,该文重点研究定点运算和有限字长存储引入的量化噪声。根据chirpscaling (CS)成像算法,建立了CS算法定点运算的量化误差模型,分析了处理流程中的量化噪声,推导了系统输出噪信比与系统字长、FFT长度等参数之间的关系。采用不同系统字长对Radarsat-I数据成像,图像质量分析与所述理论一致,结果表明:通过计算处理流程的噪信比,可实现定点SAR成像处理器系统字长等关键参数的设计。
In order to realize the miniaturization and real-time signal processing of synthetic apertureradar (SAR) imaging processor, this paper focuses on the quantization noise introduced by fixed-point arithmetic and limited word length storage. According to the chirpscaling (CS) imaging algorithm, a quantization error model of fixed-point operation in CS algorithm is established. The quantization noise in the processing flow is analyzed. The relationship between the output noise ratio of the system and parameters such as system word length and FFT length is deduced. Radarsat-I data is imaged using different system word lengths. The image quality analysis is consistent with the theory. The results show that the design of key parameters such as word length of fixed-point SAR imaging processor system can be achieved by calculating the noise-to-signal ratio of processing flow.