svm遥感图像 python

使用SVM进行遥感图像分类的Python代码示例：

1. 导入必要的库

import numpy as np
from sklearn import svm
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
from osgeo import gdal

2. 读取遥感图像数据

def read_img(filename):
    dataset = gdal.Open(filename)
    im_width = dataset.RasterXSize
    im_height = dataset.RasterYSize
    im_geotrans = dataset.GetGeoTransform()
    im_proj = dataset.GetProjection()
    im_data = dataset.ReadAsArray(0, 0, im_width, im_height)
    im_data = np.transpose(im_data, (1, 2, 0))
    im_band = im_data.reshape(im_width * im_height, -1)
    return im_band, im_data, im_width, im_height, im_geotrans, im_proj

3. 数据预处理

def preprocessing(X, y):
    # 标准化
    scaler = StandardScaler().fit(X)
    X_std = scaler.transform(X)

    # PCA降维
    pca = PCA(n_components=30).fit(X_std)
    X_pca = pca.transform(X_std)

    # 划分训练集和测试集
    X_train, X_test, y_train, y_test = train_test_split(X_pca, y, test_size=0.2, random_state=1)

    return X_train, X_test, y_train, y_test

4. 训练模型并进行预测

def train_predict(X_train, X_test, y_train, y_test):
    # SVM模型训练
    clf = svm.SVC(kernel='rbf', C=1.0, gamma='auto')
    clf.fit(X_train, y_train)

    # 预测
    y_pred = clf.predict(X_test)

    # 计算准确率
    acc = accuracy_score(y_test, y_pred)

    return acc

5. 完整代码示例

import numpy as np
from sklearn import svm
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
from osgeo import gdal

def read_img(filename):
    dataset = gdal.Open(filename)
    im_width = dataset.RasterXSize
    im_height = dataset.RasterYSize
    im_geotrans = dataset.GetGeoTransform()
    im_proj = dataset.GetProjection()
    im_data = dataset.ReadAsArray(0, 0, im_width, im_height)
    im_data = np.transpose(im_data, (1, 2, 0))
    im_band = im_data.reshape(im_width * im_height, -1)
    return im_band, im_data, im_width, im_height, im_geotrans, im_proj

def preprocessing(X, y):
    # 标准化
    scaler = StandardScaler().fit(X)
    X_std = scaler.transform(X)

    # PCA降维
    pca = PCA(n_components=30).fit(X_std)
    X_pca = pca.transform(X_std)

    # 划分训练集和测试集
    X_train, X_test, y_train, y_test = train_test_split(X_pca, y, test_size=0.2, random_state=1)

    return X_train, X_test, y_train, y_test

def train_predict(X_train, X_test, y_train, y_test):
    # SVM模型训练
    clf = svm.SVC(kernel='rbf', C=1.0, gamma='auto')
    clf.fit(X_train, y_train)

    # 预测
    y_pred = clf.predict(X_test)

    # 计算准确率
    acc = accuracy_score(y_test, y_pred)

    return acc

if __name__ == '__main__':
    # 读取数据
    filename = 'filename.tif'
    im_band, im_data, im_width, im_height, im_geotrans, im_proj = read_img(filename)

    # 获取标签
    y = np.loadtxt('labels.txt')

    # 预处理数据
    X_train, X_test, y_train, y_test = preprocessing(im_band, y)

    # 训练并预测
    acc = train_predict(X_train, X_test, y_train, y_test)

    print('Accuracy:', acc)