论文部分内容阅读
大气点扩展函数(PSF)是开展光学遥感邻近效应研究和校正的有效方法。基于蒙特卡罗模拟获得的大气PSF,设计足够多带有Sigmoid函数的隐藏神经元和线性输出神经元的两层前馈神经网络,采用LevenbergMarquardt反向传播算法,获得了大气、光谱和观测几何等输入参数与大气PSF之间的关系。模拟结果证明该方法能够在相对较短的时间内,以95%的计算精度产生预期的大气PSF的近似值。
The atmospheric point spread function (PSF) is an effective method to study and correct the proximity effect of optical remote sensing. Based on the atmospheric PSF obtained by Monte-Carlo simulation, enough two-layer feedforward neural networks with hidden neurons and linear output neurons with Sigmoid function are designed. The Levenberg-Marquardt backpropagation algorithm is used to obtain the atmospheric, spectral and observational geometry Relationship between input parameters and atmospheric PSF. Simulation results show that this method can generate the expected approximate value of PSF in 95% accuracy in a relatively short period of time.