带有高斯噪声攻击、椒盐噪声攻击、旋转攻击的基于DWT的图像信息隐藏算法代码，用python实现

以下是一个基于DWT的图像信息隐藏算法，它包括高斯噪声攻击、椒盐噪声攻击和旋转攻击的Python代码实现：

import numpy as np
import cv2
from scipy.fftpack import dct, idct
from scipy import signal
import random

def add_gaussian_noise(image, mean=0, var=0.01):
    """
    添加高斯噪声
    """
    row, col = image.shape
    sigma = var ** 0.5
    gauss = np.random.normal(mean, sigma, (row, col))
    gauss = gauss.reshape(row, col)
    noisy = image + gauss
    return noisy

def add_salt_pepper_noise(image, salt_vs_pepper=0.5, amount=0.004):
    """
    添加椒盐噪声
    """
    row, col = image.shape
    num_salt = np.ceil(amount * image.size * salt_vs_pepper)
    num_pepper = np.ceil(amount * image.size * (1.0 - salt_vs_pepper))

    # 添加椒盐噪声
    coords_salt = [np.random.randint(0, i - 1, int(num_salt)) for i in image.shape]
    image[coords_salt] = 1

    coords_pepper = [np.random.randint(0, i - 1, int(num_pepper)) for i in image.shape]
    image[coords_pepper] = 0
    return image

def rotate(image, angle=30, scale=1.0):
    """
    旋转图像
    """
    (h, w) = image.shape[:2]
    center = (w // 2, h // 2)
    M = cv2.getRotationMatrix2D(center, angle, scale)
    rotated = cv2.warpAffine(image, M, (w, h))
    return rotated

def dwt2d(img, mode='haar'):
    """
    二维离散小波变换
    """
    LL, (LH, HL, HH) = pywt.dwt2(img, mode)
    return LL, LH, HL, HH

def idwt2d(LL, LH, HL, HH, mode='haar'):
    """
    二维离散小波逆变换
    """
    img = pywt.idwt2((LL, (LH, HL, HH)), mode)
    return img

def embed(img, watermark):
    """
    图像信息隐藏
    """
    # 将水印展成一维向量
    watermark = watermark.flatten()
    # 对原始图像进行二维离散小波变换
    LL, LH, HL, HH = dwt2d(img)

    # 将LH、HL、HH系数转化为一维数组
    LH = LH.flatten()
    HL = HL.flatten()
    HH = HH.flatten()

    # 将水印嵌入LH、HL、HH系数的最后一位
    for i in range(len(watermark)):
        LH[i] = (LH[i] & 254) | watermark[i]
        HL[i] = (HL[i] & 254) | watermark[i]
        HH[i] = (HH[i] & 254) | watermark[i]

    # 将修改后的LH、HL、HH系数重新组合成矩阵
    LH = LH.reshape(LL.shape[0] // 2, LL.shape[1] // 2)
    HL = HL.reshape(LL.shape[0] // 2, LL.shape[1] // 2)
    HH = HH.reshape(LL.shape[0] // 2, LL.shape[1] // 2)

    # 将修改后的系数与LL系数组合成二维离散小波变换系数矩阵
    coeffs = (LL, (LH, HL, HH))

    # 对修改后的系数进行二维离散小波逆变换，得到嵌入水印后的图像
    watermarked_img = idwt2d(*coeffs)

    return watermarked_img

def extract(img, watermark_shape):
    """
    图像信息提取
    """
    # 对水印形状进行验证
    watermark_size = np.prod(watermark_shape)
    assert watermark_size <= img.size, '水印大小超过图像大小'

    # 对原始图像进行二维离散小波变换
    LL, LH, HL, HH = dwt2d(img)

    # 将LH、HL、HH系数转化为一维数组
    LH = LH.flatten()
    HL = HL.flatten()
    HH = HH.flatten()

    # 从LH、HL、HH系数的最后一位中提取水印
    watermark = np.zeros(watermark_shape, dtype=np.uint8)
    for i in range(watermark_size):
        watermark[i // watermark_shape[1], i % watermark_shape[1]] = LH[i] & 1

    # 返回提取的水印
    return watermark

if __name__ == '__main__':
    # 读取原始图像
    img = cv2.imread('lena.png', cv2.IMREAD_GRAYSCALE)

    # 对原始图像进行预处理，添加高斯噪声、椒盐噪声、旋转攻击
    img_noise_gaussian = add_gaussian_noise(img)
    img_noise_salt_pepper = add_salt_pepper_noise(img)
    img_rotate = rotate(img, angle=30, scale=1.0)

    # 显示原始图像和预处理后的图像
    cv2.imshow('Original', img)
    cv2.imshow('Gaussian noise', img_noise_gaussian)
    cv2.imshow('Salt & pepper noise', img_noise_salt_pepper)
    cv2.imshow('Rotate', img_rotate)

    # 嵌入水印
    watermark = np.random.randint(0, 2, (128, 128), dtype=np.uint8)
    watermarked_img = embed(img, watermark)
    watermarked_img_noise_gaussian = embed(img_noise_gaussian, watermark)
    watermarked_img_noise_salt_pepper = embed(img_noise_salt_pepper, watermark)
    watermarked_img_rotate = embed(img_rotate, watermark)

    # 提取水印
    extracted_watermark = extract(watermarked_img, watermark_shape=watermark.shape)
    extracted_watermark_noise_gaussian = extract(watermarked_img_noise_gaussian, watermark_shape=watermark.shape)
    extracted_watermark_noise_salt_pepper = extract(watermarked_img_noise_salt_pepper, watermark_shape=watermark.shape)
    extracted_watermark_rotate = extract(watermarked_img_rotate, watermark_shape=watermark.shape)

    # 对比原始水印和提取出的水印
    print('原始水印：\n', watermark)
    print('提取出的水印（原始图像）：\n', extracted_watermark)
    print('提取出的水印（高斯噪声图像）：\n', extracted_watermark_noise_gaussian)
    print('提取出的水印（椒盐噪声图像）：\n', extracted_watermark_noise_salt_pepper)
    print('提取出的水印（旋转攻击图像）：\n', extracted_watermark_rotate)

    # 显示嵌入水印后的图像和提取出的水印
    cv2.imshow('Watermarked (Original)', watermarked_img)
    cv2.imshow('Watermarked (Gaussian noise)', watermarked_img_noise_gaussian)
    cv2.imshow('Watermarked (Salt & pepper noise)', watermarked_img_noise_salt_pepper)
    cv2.imshow('Watermarked (Rotate)', watermarked_img_rotate)
    cv2.imshow('Extracted watermark (Original)', extracted_watermark * 255)
    cv2.imshow('Extracted watermark (Gaussian noise)', extracted_watermark_noise_gaussian * 255)
    cv2.imshow('Extracted watermark (Salt & pepper noise)', extracted_watermark_noise_salt_pepper * 255)
    cv2.imshow('Extracted watermark (Rotate)', extracted_watermark_rotate * 255)

    cv2.waitKey(0)
    cv2.destroyAllWindows()

在这个代码中，我们实现了一个基于DWT的图像信息隐藏算法，并对原始图像进行了高斯噪声攻击、椒盐噪声攻击和旋转攻击。我们首先读取原始图像，然后添加三种类型的攻击。我们生成一个随机水印，并将其嵌入到原始图像、高斯噪声图像、椒盐噪声图像和旋转攻击图像中。最后，我们提取出嵌入的水印，并将其与原始水印进行比较。