请帮我写一段代码，生成对抗神经网络对图像降噪

生成对抗网络（GANs）用于图像处理的场景下，通常包括一个生成器（Generator）和一个判别器（Discriminator）。对于图像降噪任务，我们可以创建一个简单的卷积神经网络作为基础模型。这里是一个使用Keras库（Python）实现的简版示例：

import tensorflow as tf
from tensorflow.keras.layers import Input, Conv2D, Reshape, Dense, Flatten, LeakyReLU, Dropout
from tensorflow.keras.models import Model

# 假设我们有8x8的二维噪声输入 (batch_size, 8, 8, noise_dim)
input_shape = (None, 8, 8, 100) # noise_dim可以根据需要调整
noise_input = Input(shape=input_shape, name='input_noise')

# 定义生成器部分
def build_generator():
    x = Conv2DTranspose(64, (4, 4), strides=2, padding='same', use_bias=False)(noise_input)
    x = BatchNormalization()(x)
    x = LeakyReLU(alpha=0.2)(x)

    x = Conv2DTranspose(128, (4, 4), strides=2, padding='same', use_bias=False)(x)
    x = BatchNormalization()(x)
    x = LeakyReLU(alpha=0.2)(x)

    x = Conv2DTranspose(256, (4, 4), strides=2, padding='same', use_bias=False)(x)
    x = BatchNormalization()(x)
    x = LeakyReLU(alpha=0.2)(x)

    x = Conv2DTranspose(1, (4, 4), strides=2, activation='tanh', padding='same')(x)
    return x

generator = build_generator()
gen_output = generator(noise_input)

# 定义判别器部分
def build_discriminator():
    x = Conv2D(64, (4, 4), strides=2, padding='same', use_bias=False)(gen_output)
    x = LeakyReLU(alpha=0.2)(x)
    x = Dropout(0.25)(x)

    x = Conv2D(128, (4, 4), strides=2, padding='same', use_bias=False)(x)
    x = BatchNormalization()(x)
    x = LeakyReLU(alpha=0.2)(x)
    x = Dropout(0.25)(x)

    x = Conv2D(256, (4, 4), strides=2, padding='same', use_bias=False)(x)
    x = BatchNormalization()(x)
    x = LeakyReLU(alpha=0.2)(x)
    x = Dropout(0.25)(x)

    x = Flatten()(x)
    x = Dense(1, activation='sigmoid')(x)  # 输出单通道二值结果
    return Model(inputs=noise_input, outputs=x)

discriminator = build_discriminator()

# GAN整体结构
gan_model = Model(inputs=noise_input, outputs=[gen_output, discriminator(gen_output)])

# 编译GAN模型（假设我们用Adam优化器，损失函数可以是Wasserstein Loss和Gradient Penalty）
gan_model.compile(optimizer=tf.optimizers.Adam(), loss=['mae', 'binary_crossentropy'])