遥感图像光谱特征，纹理特征，树高特征融合代码

遥感图像的光谱特征、纹理特征和树高特征融合可以通过机器学习算法来实现。下面给出一份基于Python的代码示例，使用随机森林算法进行特征融合。

首先，需要安装必要的库，包括numpy、pandas、sklearn、gdal等：

import numpy as np
import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from osgeo import gdal

接下来，读取遥感图像数据，获取光谱特征和纹理特征：

def get_spectral_texture_features(img_path):
    # 读取遥感图像
    src_ds = gdal.Open(img_path)
    img = src_ds.ReadAsArray()
    
    # 计算光谱特征
    spectral_features = np.zeros((img.shape[1], img.shape[2], img.shape[0]))
    for i in range(img.shape[0]):
        spectral_features[:, :, i] = img[i, :, :]
    
    # 计算纹理特征
    texture_features = np.zeros((img.shape[1], img.shape[2], img.shape[0]))
    for i in range(img.shape[0]):
        glcm = greycomatrix(img[i, :, :], [1], [0, np.pi/4, np.pi/2, 3*np.pi/4], levels=256, symmetric=True, normed=True)
        texture_features[:, :, i] = np.concatenate([greycoprops(glcm, 'contrast'),
                                                     greycoprops(glcm, 'dissimilarity'),
                                                     greycoprops(glcm, 'homogeneity'),
                                                     greycoprops(glcm, 'energy'),
                                                     greycoprops(glcm, 'correlation')], axis=1)
    
    return spectral_features, texture_features

然后，读取树高数据，获取树高特征：

def get_height_feature(height_path):
    # 读取树高数据
    height_ds = gdal.Open(height_path)
    height = height_ds.ReadAsArray()
    
    return height.reshape(-1, 1)

接下来，定义一个函数，将光谱特征、纹理特征和树高特征融合：

def fusion_features(spectral_features, texture_features, height_feature):
    # 将光谱特征和纹理特征合并
    features = np.concatenate([spectral_features.reshape(-1, spectral_features.shape[2]),
                                texture_features.reshape(-1, texture_features.shape[2])], axis=1)
    
    # 将树高特征加入到特征矩阵中
    features = np.concatenate([features, height_feature], axis=1)
    
    return features

最后，使用随机森林算法对融合后的特征进行分类：

def classify(features, labels):
    # 随机森林分类器
    rf = RandomForestClassifier(n_estimators=100, max_depth=10)
    rf.fit(features, labels)
    
    return rf.predict(features)

完整的代码如下：

import numpy as np
import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from osgeo import gdal
from skimage.feature import greycomatrix, greycoprops

def get_spectral_texture_features(img_path):
    # 读取遥感图像
    src_ds = gdal.Open(img_path)
    img = src_ds.ReadAsArray()
    
    # 计算光谱特征
    spectral_features = np.zeros((img.shape[1], img.shape[2], img.shape[0]))
    for i in range(img.shape[0]):
        spectral_features[:, :, i] = img[i, :, :]
    
    # 计算纹理特征
    texture_features = np.zeros((img.shape[1], img.shape[2], img.shape[0]))
    for i in range(img.shape[0]):
        glcm = greycomatrix(img[i, :, :], [1], [0, np.pi/4, np.pi/2, 3*np.pi/4], levels=256, symmetric=True, normed=True)
        texture_features[:, :, i] = np.concatenate([greycoprops(glcm, 'contrast'),
                                                     greycoprops(glcm, 'dissimilarity'),
                                                     greycoprops(glcm, 'homogeneity'),
                                                     greycoprops(glcm, 'energy'),
                                                     greycoprops(glcm, 'correlation')], axis=1)
    
    return spectral_features, texture_features

def get_height_feature(height_path):
    # 读取树高数据
    height_ds = gdal.Open(height_path)
    height = height_ds.ReadAsArray()
    
    return height.reshape(-1, 1)

def fusion_features(spectral_features, texture_features, height_feature):
    # 将光谱特征和纹理特征合并
    features = np.concatenate([spectral_features.reshape(-1, spectral_features.shape[2]),
                                texture_features.reshape(-1, texture_features.shape[2])], axis=1)
    
    # 将树高特征加入到特征矩阵中
    features = np.concatenate([features, height_feature], axis=1)
    
    return features

def classify(features, labels):
    # 随机森林分类器
    rf = RandomForestClassifier(n_estimators=100, max_depth=10)
    rf.fit(features, labels)
    
    return rf.predict(features)

if __name__ == '__main__':
    # 获取光谱特征和纹理特征
    spectral_features, texture_features = get_spectral_texture_features('image.tif')
    
    # 获取树高特征
    height_feature = get_height_feature('height.tif')
    
    # 特征融合
    features = fusion_features(spectral_features, texture_features, height_feature)
    
    # 进行分类
    labels = np.loadtxt('labels.txt', delimiter=',')
    predictions = classify(features, labels)
    
    # 输出分类结果
    print(predictions)