一些基础的图像增强方法，批量处理饱和度、噪声、对比度等调整

帮助你高效处理图片，扩大训练集数量，提高训练精度。

1.对比度、锐度、亮度

from PIL import Image
from PIL import ImageEnhance
import os

rootdir = r'/media/jidong/code/zdata/val'  # 指明被遍历的文件夹
for parent, dirnames, filenames in os.walk(rootdir):
    for filename in filenames:
        currentPath = os.path.join(parent, filename)
        im = Image.open(currentPath)
        print(filename)
        # enh_con = ImageEnhance.Contrast(im)  # 增加对比度
        # image_contrasted = enh_con.enhance(0.5)
        # newname1 = r"/media/jidong/code/zdata/tr-duib/" + filename
        # image_contrasted.save(newname1)

        # enh_sha = ImageEnhance.Sharpness(im)  # 增加锐度
        # image_sharped = enh_sha.enhance(3.0)
        # newname2 = r"/media/jidong/code/zdata/vl-sha/" + filename
        # image_sharped.save(newname2)

        # 增加亮度
        enh_bri = ImageEnhance.Brightness(im)
        image_brightened = enh_bri.enhance(0.5)
        newname3 = r"/media/jidong/code/zdata/vl-brig/" + filename
        image_brightened.save(newname3)
        #
        # enh_col = ImageEnhance.Color(im_rotate)  # 增加色度 但是有问题,
        # image_colored = enh_col.enhance(1.5)
        # newname4 = r"/home/qxq/Desktop/eyedata_final/mask/brighten/" + 'Col_' + filename
        # image_contrasted.save(newname4)

        j = 0

2.椒盐噪声、高斯噪声

import os
import cv2
import numpy as np
import random


def sp_noise(noise_img, proportion):
    '''
    添加椒盐噪声
    proportion的值表示加入噪声的量，可根据需要自行调整
    return: img_noise
    '''
    height, width = noise_img.shape[0], noise_img.shape[1]#获取高度宽度像素值
    num = int(height * width * proportion) #一个准备加入多少噪声小点
    for i in range(num):
        w = random.randint(0, width - 1)
        h = random.randint(0, height - 1)
        if random.randint(0, 1) == 0:
            noise_img[h, w] = 0
        else:
            noise_img[h, w] = 255
    return noise_img


def gaussian_noise(img, mean, sigma):
    '''
    此函数用将产生的高斯噪声加到图片上
    传入:
        img   :  原图
        mean  :  均值
        sigma :  标准差
    返回:
        gaussian_out : 噪声处理后的图片
    '''
    # 将图片灰度标准化
    img = img / 255
    # 产生高斯 noise
    noise = np.random.normal(mean, sigma, img.shape)
    # 将噪声和图片叠加
    gaussian_out = img + noise
    # 将超过 1 的置 1，低于 0 的置 0
    gaussian_out = np.clip(gaussian_out, 0, 1)
    # 将图片灰度范围的恢复为 0-255
    gaussian_out = np.uint8(gaussian_out*255)
    # 将噪声范围搞为 0-255
    # noise = np.uint8(noise*255)
    return gaussian_out# 这里也会返回噪声，注意返回值


def convert(input_dir, output_dir):
    for filename in os.listdir(input_dir):
        path = input_dir + "/" + filename  # 获取文件路径
        print("doing... ", path)
        noise_img = cv2.imread(path)  # 读取图片
        # img_noise = gaussian_noise(noise_img, 0, 0.1)  # 高斯噪声
        img_noise = sp_noise(noise_img, 0.05)# 椒盐噪声
        # img_noise  = random_noise(noise_img,500)# 随机噪声
        cv2.imwrite(output_dir+'/'+filename, img_noise)


if __name__ == '__main__':
    input_dir = "/media/jidong/code/zdata/val"    # 输入数据文件夹
    output_dir = "/media/jidong/code/zdata/vl-jzao"   # 输出数据文件夹
    convert(input_dir, output_dir)

3.翻转

# -*- coding: UTF-8 -*-

from PIL import Image
import os

# 获得文件夹下所有文件
filePath = '/media/jidong/code/shilidata/val/'
filenames = os.listdir(filePath)

# 指定保存的文件夹
outputPath = '/media/jidong/code/shilidata/vl-fan/'

# 迭代所有图片
for filename in filenames:
    # 读取图像
    im = Image.open(filePath + filename)

    # 指定逆时针旋转的角度
    im_rotate = im.rotate(180)

    # 保存图像
    im_rotate.save(outputPath + filename)

4.饱和度

import numpy as np
import cv2
import os

# 调整最大值
MAX_VALUE = 100


def update(input_img_path, output_img_path, lightness, saturation):
    """
    用于修改图片的亮度和饱和度
    :param input_img_path: 图片路径
    :param output_img_path: 输出图片路径
    :param lightness: 亮度
    :param saturation: 饱和度
    """

    # 加载图片 读取彩色图像归一化且转换为浮点型
    image = cv2.imread(input_img_path, cv2.IMREAD_COLOR).astype(np.float32) / 255.0

    # 颜色空间转换 BGR转为HLS
    hlsImg = cv2.cvtColor(image, cv2.COLOR_BGR2HLS)

    # 1.调整亮度（线性变换)
    hlsImg[:, :, 1] = (1.0 + lightness / float(MAX_VALUE)) * hlsImg[:, :, 1]
    hlsImg[:, :, 1][hlsImg[:, :, 1] > 1] = 1
    # 饱和度
    hlsImg[:, :, 2] = (1.0 + saturation / float(MAX_VALUE)) * hlsImg[:, :, 2]
    hlsImg[:, :, 2][hlsImg[:, :, 2] > 1] = 1
    # HLS2BGR
    lsImg = cv2.cvtColor(hlsImg, cv2.COLOR_HLS2BGR) * 255
    lsImg = lsImg.astype(np.uint8)
    cv2.imwrite(output_img_path, lsImg)


dataset_dir = '/media/jidong/code/zdata/val'
output_dir = '/media/jidong/code/zdata/vl-baoh'

# 这里调参！！！
lightness = int(input("lightness(亮度-100~+100):"))  # 亮度
saturation = int(input("saturation(饱和度-100~+100):"))  # 饱和度

# 获得需要转化的图片路径并生成目标路径
image_filenames = [(os.path.join(dataset_dir, x), os.path.join(output_dir, x))
                   for x in os.listdir(dataset_dir)]
# 转化所有图片
for path in image_filenames:
    update(path[0], path[1], lightness, saturation)

祝你早日发CVPR、ICCV等一流会议、期刊

加油，加油，加油！！！！！