设计一个对MNIST处理的CNN网络，网络结构和参数要求如下表，其中每一次的MaxPooling后都添加系数为0.25的Dropout的完整python代码

import keras
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Dropout, Flatten, Dense

# define the input shape
input_shape = (28, 28, 1)

# define the number of filters
filters = [32, 64, 128]

# define the kernel size
kernel_sizes = [3, 3, 3]

# define the pooling size
pool_sizes = [2, 2, 2]

# define the dropout rate
dropout_rate = 0.25

# define the number of classes
num_classes = 10

# create a sequential model
model = Sequential()

# add the first convolutional layer
model.add(Conv2D(filters[0], kernel_sizes[0], activation='relu', input_shape=input_shape))
model.add(MaxPooling2D(pool_sizes[0]))
model.add(Dropout(dropout_rate))

# add the second convolutional layer
model.add(Conv2D(filters[1], kernel_sizes[1], activation='relu'))
model.add(MaxPooling2D(pool_sizes[1]))
model.add(Dropout(dropout_rate))

# add the third convolutional layer
model.add(Conv2D(filters[2], kernel_sizes[2], activation='relu'))
model.add(MaxPooling2D(pool_sizes[2]))
model.add(Dropout(dropout_rate))

# flatten the feature maps
model.add(Flatten())

# add a fully connected layer
model.add(Dense(256, activation='relu'))
model.add(Dropout(dropout_rate))

# add the output layer
model.add(Dense(num_classes, activation='softmax'))

# compile the model
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])