利用波谱分析进行SPOT影像自然色模拟的方法
SPOT Natural-Color Simulation with Spectrum Analysis
-
摘要: 常规线性合成方法对波段之间关系的描述过于简单,难以产生高度真实感的SPOT自然色影像。针对这个问题,提出了利用波谱分析的SPOT自然色模拟方法。该方法解决了波谱库波谱与像元波谱的尺度转换问题,利用BP神经网络对波谱库中的地物波谱进行学习和分析,拟合波谱库波段间的非线性关系,然后利用这样的非线性关系模拟SPOT蓝波段,从而得到SPOT近自然色影像。实验结果表明,该方法得到的SPOT近自然色影像无论是视觉效果还是波谱信息均优于常规的波段合成方法。Abstract: Traditional linear models for generation SPOT natural-color is difficult to produce highly realistic natural-color SPOT image because of its simplified description of the relationship between the image bands.To overcome this problem,we propose a new SPOT natural-color simulation method based on spectrum analysis.Firstly,we take the image's spectrum as a reference to adjust spectrum library's spectrum scales and form some spectrum bands.Then,we use the back-propagation artificial neural network(BP ANN) to study and analyze the relationship between spectrum-library's blue band and other bands.Finally,we take this relationship into the simulation of SPOT blue band and generate the natural-color images.Experiment show that the natural-color image is much more excellent in both visual effects and spectrum information than others.
下载:
