论文部分内容阅读
采用盲源分离方法解混岩矿混合像元,获取岩矿组分信息.分析了常用的Fast ICA、Robust ICA方法,从算法稳定性、分离信号质量和迭代计算效率3个方面,比较不同目标函数及寻优过程的优势和不足:Robust ICA在算法稳定性和迭代计算效率上具有较大优势,分离信号质量并不是最佳;Fast ICA对初始值和步长比较敏感,计算可能不收敛,也可能陷于局部最优;峭度为目标函数的Fast ICA有较好的分离信号质量,但算法稳定性不如负熵为目标函数的Fast ICA;用于负熵近似的非二次函数对算法稳定性和迭代计算效率有较大影响,原因是非二次函数影响迭代计算步长,较小的步长算法稳定性较好,但是迭代计算效率降低.实际运用中,应根据岩矿混合像元光谱特点,选择恰当的混合像元分离方法,在不同性能之间达到平衡.
The blind source separation method is used to disaggregate mixed pixel of rock and mine to obtain the information of rock and mine components.It analyzes the commonly used Fast ICA and Robust ICA methods and compares different targets from three aspects of algorithm stability, signal quality separation and iterative calculation efficiency Function and optimization process: Robust ICA has a great advantage in algorithm stability and iterative computing efficiency, signal quality is not the best separation; Fast ICA is sensitive to the initial value and step size, the calculation may not converge, But also may fall into local optimum. Fast ICA with kurtosis as the objective function has better signal quality, but the stability of the algorithm is not as good as Fast ICA with negative entropy as the objective function. Non-quadratic function used for negative entropy approximation is stable to the algorithm The reason is that the non-quadratic function affects the iterative calculation step, the smaller step algorithm has better stability, but the iterative calculation efficiency is reduced. In practice, according to the rock-mine mixed pixel spectrum Features, choose the appropriate hybrid pixel separation method, to achieve a balance between different properties.