批处理对图像数据进行数据增强

# -*- coding: utf-8 -*-

import cv2

import numpy as np

import os.path

import copy

# 椒盐噪声

def SaltAndPepper(src,percetage):

    SP_NoiseImg=src.copy()

    SP_NoiseNum=int(percetage*src.shape[0]*src.shape[1])

    for i in range(SP_NoiseNum):

        randR=np.random.randint(0,src.shape[0]-1)

        randG=np.random.randint(0,src.shape[1]-1)

        randB=np.random.randint(0,3)

        if np.random.randint(0,1)==0:

            SP_NoiseImg[randR,randG,randB]=0

        else:

            SP_NoiseImg[randR,randG,randB]=255

    return SP_NoiseImg

# 高斯噪声

def addGaussianNoise(image,percetage):

    G_Noiseimg = image.copy()

    w = image.shape[1]

    h = image.shape[0]

    G_NoiseNum=int(percetage*image.shape[0]*image.shape[1])

    for i in range(G_NoiseNum):

        temp_x = np.random.randint(0,h)

        temp_y = np.random.randint(0,w)

        G_Noiseimg[temp_x][temp_y][np.random.randint(3)] = np.random.randn(1)[0]

    return G_Noiseimg

# 昏暗

def darker(image,percetage=0.9):

    image_copy = image.copy()

    w = image.shape[1]

    h = image.shape[0]

    #get darker

    for xi in range(0,w):

        for xj in range(0,h):

            image_copy[xj,xi,0] = int(image[xj,xi,0]*percetage)

            image_copy[xj,xi,1] = int(image[xj,xi,1]*percetage)

            image_copy[xj,xi,2] = int(image[xj,xi,2]*percetage)

    return image_copy

# 亮度

def brighter(image, percetage=1.5):

    image_copy = image.copy()

    w = image.shape[1]

    h = image.shape[0]

    #get brighter

    for xi in range(0,w):

        for xj in range(0,h):

            image_copy[xj,xi,0] = np.clip(int(image[xj,xi,0]*percetage),a_max=255,a_min=0)

            image_copy[xj,xi,1] = np.clip(int(image[xj,xi,1]*percetage),a_max=255,a_min=0)

            image_copy[xj,xi,2] = np.clip(int(image[xj,xi,2]*percetage),a_max=255,a_min=0)

    return image_copy

# 旋转

def rotate(image, angle, center=None, scale=1.0):

    (h, w) = image.shape[:2]

    # If no rotation center is specified, the center of the image is set as the rotation center

    if center is None:

        center = (w / 2, h / 2)

    m = cv2.getRotationMatrix2D(center, angle, scale)

    rotated = cv2.warpAffine(image, m, (w, h))

    return rotated

# 翻转

def flip(image):

    flipped_image = np.fliplr(image)

    return flipped_image

 

from PIL import Image, ImageEnhance

import os

import random

import shutil

def augment_image(image_path, save_path):

    img = cv2.imread(image_path)

    image_name = os.path.basename(image_path)  # 获取图片名称

    split_result = image_name.split('.')

    name = split_result[:-1]

    extension = split_result[-1]

    # cv2.imshow("1",img)

    # cv2.waitKey(5000)

    # 旋转

    rotated_90 = rotate(img, 90)

    cv2.imwrite(save_path +  "".join(name) + '_r90.'+ extension, rotated_90)

    rotated_180 = rotate(img, 180)

    cv2.imwrite(save_path +  "".join(name) + '_r180.'+ extension, rotated_180)

    flipped_img = flip(img)

    cv2.imwrite(save_path +  "".join(name) + '_fli.'+ extension, flipped_img)

    # 增加噪声

    # img_salt = SaltAndPepper(img, 0.3)

    # cv2.imwrite(save_path + img_name[0:7] + '_salt.jpg', img_salt)

    img_gauss = addGaussianNoise(img, 0.3)

    cv2.imwrite(save_path +  "".join(name) + '_noise.'+ extension,img_gauss)

    #变亮、变暗

    img_darker = darker(img)

    cv2.imwrite(save_path +  "".join(name) + '_darker.'+ extension, img_darker)

    img_brighter = brighter(img)

    cv2.imwrite(save_path +  "".join(name) + '_brighter.'+ extension, img_brighter)

    blur = cv2.GaussianBlur(img, (7, 7), 1.5)

    #      cv2.GaussianBlur(图像，卷积核，标准差）

    cv2.imwrite(save_path +  "".join(name) + '_blur.'+ extension,blur)

target_num = 2000  # 目标增强图片数量

image_folder = 'D:/video/bk/img/'  # 图片文件夹路径

save_folder = 'D:/video/bk/img/augmentation/'  # 保存增强后的图片的文件夹路径

 

if not os.path.exists(save_folder): #判断所在目录下是否有该文件名的文件夹

    os.mkdir(save_folder) #创建多级目录用mkdirs，单击目录mkdir

# 获取所有类别的文件夹路径

class_folders = os.listdir(image_folder)

for root, dirs, files in os.walk(image_folder):  # 这里就填文件夹目录就可以了

    for file in files:

        # 获取文件路径

        if ('.jpg' in file):

            image_path =  image_folder+ file

            target_path = save_folder

            augment_image(image_path, target_path)