#代碼中的noisef為信號(hào)等級(jí)，例如我需要0.7的噪聲，傳入?yún)?shù)我傳入的是1-0.7
from PIL import Image
import numpy as np
import random

import torchvision.transforms as transforms

norm_mean = (0.5, 0.5, 0.5)
norm_std = (0.5, 0.5, 0.5)
class AddPepperNoise(object):
    """增加椒鹽噪聲
    Args:
        snr （float）: Signal Noise Rate
        p (float): 概率值，依概率執(zhí)行該操作
    """

    def __init__(self, snr, p=0.9):
        assert isinstance(snr, float) and (isinstance(p, float))    # 2020 07 26 or --> and
        self.snr = snr
        self.p = p

    def __call__(self, img):
        """
        Args:
            img (PIL Image): PIL Image
        Returns:
            PIL Image: PIL image.
        """
        if random.uniform(0, 1) < self.p:
            img_ = np.array(img).copy()
            h, w, c = img_.shape
            signal_pct = self.snr
            noise_pct = (1 - self.snr)
            mask = np.random.choice((0, 1, 2), size=(h, w, 1), p=[signal_pct, noise_pct/2., noise_pct/2.])
            mask = np.repeat(mask, c, axis=2)
            img_[mask == 1] = 255   # 鹽噪聲
            img_[mask == 2] = 0     # 椒噪聲
            return Image.fromarray(img_.astype('uint8')).convert('RGB')
        else:
            return img

class Gaussian_noise(object):
    """增加高斯噪聲
    此函數(shù)用將產(chǎn)生的高斯噪聲加到圖片上
    傳入:
        img   :  原圖
        mean  :  均值
        sigma :  標(biāo)準(zhǔn)差
    返回:
        gaussian_out : 噪聲處理后的圖片
    """

    def __init__(self, mean, sigma):

        self.mean = mean
        self.sigma = sigma

    def __call__(self, img):
        """
        Args:
            img (PIL Image): PIL Image
        Returns:
            PIL Image: PIL image.
        """
        # 將圖片灰度標(biāo)準(zhǔn)化
        img_ = np.array(img).copy()
        img_ = img_ / 255.0
        # 產(chǎn)生高斯 noise
        noise = np.random.normal(self.mean, self.sigma, img_.shape)
        # 將噪聲和圖片疊加
        gaussian_out = img_ + noise
        # 將超過(guò) 1 的置 1，低于 0 的置 0
        gaussian_out = np.clip(gaussian_out, 0, 1)
        # 將圖片灰度范圍的恢復(fù)為 0-255
        gaussian_out = np.uint8(gaussian_out*255)
        # 將噪聲范圍搞為 0-255
        # noise = np.uint8(noise*255)
        return Image.fromarray(gaussian_out).convert('RGB')

def image_transform(noisef):
    """對(duì)訓(xùn)練集和測(cè)試集的圖片作預(yù)處理轉(zhuǎn)換
        train_transform：加噪圖
        _train_transform：原圖（不加噪）
        test_transform：測(cè)試圖（不加噪）
    """
    train_transform = transforms.Compose([
        transforms.Resize((256, 256)),  # 重設(shè)大小
        #transforms.RandomCrop(32,padding=4),
        AddPepperNoise(noisef, p=0.9),                 #加椒鹽噪聲

        #Gaussian_noise(0, noisef),  # 加高斯噪聲

        transforms.ToTensor(),  # 轉(zhuǎn)換為張量
        # transforms.Normalize(norm_mean,norm_std),
    ])
    _train_transform = transforms.Compose([
        transforms.Resize((256, 256)),
        #transforms.RandomCrop(32,padding=4),
        transforms.ToTensor(),
        # transforms.Normalize(norm_mean,norm_std),

    ])
    test_transform = transforms.Compose([
        transforms.Resize((256, 256)),
        #transforms.RandomCrop(32,padding=4),
        transforms.ToTensor(),
        # transforms.Normalize(norm_mean,norm_std),

    ])
    return train_transform, _train_transform, test_transform

# 圖像變換和加噪聲train_transform為加噪圖，_train_transform為原圖，test_transform為測(cè)試圖   noisef為傳入的噪聲等級(jí)
train_transform,_train_transform,test_transform = image_transform(noisef)

training_data=FabricDataset_file(data_dir=train_dir,transform=train_transform)
_training_data=FabricDataset_file(data_dir=_train_dir,transform=_train_transform)
testing_data=FabricDataset_file(data_dir=test_dir,transform=test_transform) 

def random_noise(image,noise_num):
    '''
    添加隨機(jī)噪點(diǎn)（實(shí)際上就是隨機(jī)在圖像上將像素點(diǎn)的灰度值變?yōu)?55即白色）
    :param image: 需要加噪的圖片
    :param noise_num: 添加的噪音點(diǎn)數(shù)目，一般是上千級(jí)別的
    :return: img_noise
    '''
    #
    # 參數(shù)image：，noise_num：
    img = cv2.imread(image)
    img_noise = img
    # cv2.imshow("src", img)
    rows, cols, chn = img_noise.shape
    # 加噪聲
    for i in range(noise_num):
        x = np.random.randint(0, rows)#隨機(jī)生成指定范圍的整數(shù)
        y = np.random.randint(0, cols)
        img_noise[x, y, :] = 255
    return img_noise

img_noise = random_noise("colorful_lena.jpg",3000)
cv2.imshow('random_noise',img_noise)
cv2.waitKey(0)