提出一种基于分形理论和BP 神经网络的航空遥感图像有监督分类方法。该方法尝试将航空图像

的光谱信息和纹理特征相结合。它首先将彩色航空图像由RGB 格式转化为HSI 格式,然后,根据亮度计算分

数维、多重分形广义维数谱q-D( q) 和“空隙”等基于分形的纹理特征,同时加入归一化的色度和饱和度作为光

谱特征,采用BP 神经网络作为分类器。通过对彩色航空图像的分类实验,结果证实该方法行之有效。-Based on fractal theory and BP neural network of aviation remote sensing image supervised classification method. This method tries to aerial images of the spectral information and texture characteristics of the combination. It will first color aerial images from the RGB format into HSI format, and then, according to the brightness calculation of fractal dimension, the generalized multi-fractal dimension spectrum qD (q) and the