keras实现数据的增多

在数量量少的情况下，我们需要增多数据

 

首先如果不想太大改变数据的原有属性，可以通过景象和噪音，对比度来增加数据。代码如下

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

 

from PIL import Image

import random
import tensorflow as tf
import os

import numpy as np

import argparse


 

#读取文件内的所有,迭代，将生成的图片与原图片放在同一文件夹下

 

def read_file_all(data_dir_path):
    print("在中国",data_dir_path)
    for f in os.listdir(data_dir_path):

        print("data_dir_path:",data_dir_path)

        data_file_path = os.path.join(data_dir_path, f)

        if os.path.isfile(data_file_path):

            image_rotate(data_file_path, data_dir_path)

			

            #print(collected)

        else:

            read_file_all(data_file_path)

            #文件夹下套文件夹情况

 

def read_file_all_again(data_dir_path):

	for f in os.listdir(data_dir_path):

		print("data_dir_path:",data_dir_path)

		data_file_path = os.path.join(data_dir_path, f)

		if os.path.isfile(data_file_path):

			relight(data_file_path, data_dir_path, random.uniform(0.5, 1.5), random.randint(-50, 50))

		else:

			read_file_all_again(data_file_path)

            #文件夹下套文件夹情况

					

# 改变图片的亮度与对比度

def relight(image_path, save_dir, light=1, bias=0):
   img = Image.open(image_path)
   img = np.array(img)
   w = img.shape[1]
   h = img.shape[0]
   for i in range(w):
      for j in range(h):
          for c in range(3):  #三通道
              tmp = int(img[j, i, c]*light + bias)
              if tmp > 255:
                 tmp = 255
              elif tmp < 0:
                 tmp = 0
              img[j, i, c] = tmp
   img = Image.fromarray(img)
   save_path = save_dir + "/" + random_name() + '.bmp'
   img.save(save_path)
   print("save_path:",save_path)

 

def image_rotate(image_path,save_dir):

	#读取图像

	im = Image.open(image_path)

	im = im.transpose(Image.FLIP_LEFT_RIGHT)#左右互换

	save_path = save_dir + "/" + random_name() + '.jpg'

	im.save(save_path)

	print("save_path:",save_path)

	# img=np.array(Image.open(image_path)) 

    # #随机生成100个椒盐

    # rows,cols,dims=img.shape

    # for i in range(100):

        # x=np.random.randint(0,rows)

        # y=np.random.randint(0,cols)

        # img[x,y,:]=255

    # img.flags.writeable = True  # 将数组改为读写模式

    # dst=Image.fromarray(np.uint8(img))

    # save_path = save_dir + "/" + random_name() + '.bmp'

    # dst.save(save_path)

    # print("save_path:",save_path)

def random_name():

    #随机数，用来随机取名字

	a_list = ['0','1','2','3','4','5','6','7','8','9']

	name = random.sample(a_list,5)

	file_name = "".join(name)

	return file_name

 

# construct the argument parser and parse the arguments


data_dir_path = "E:/glass_class_dectect/data/细框眼镜599"#读取/保存的文件路径

read_file_all(data_dir_path)
#read_file_all_again(data_dir_path)

如果不考虑旋转，增大图像的影响。我们也可以通过起扩大我们的数据集，代码如下

 

# -*- coding: utf-8 -*-
"""
Created on Fri Aug  2 17:06:02 2019

@author: 01
"""

import skimage.io as io
import os,sys
from skimage import data_dir
import numpy as np

from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img

datagen = ImageDataGenerator(
        rotation_range=0,
        width_shift_range=0.2,
        height_shift_range=0.2,
        shear_range=0.2,
        zoom_range=0.2,
        horizontal_flip=True,
        fill_mode='nearest')
path="E:/glass_class_dectect/data/无框眼镜127/";     
dirs=os.listdir(path)
x_all=[]
for file in dirs:
    img = load_img(path+file)
    x = img_to_array(img)  
    x = x.reshape((1,) + x.shape)  # 这是一个numpy数组，形状为 (1, 3, 150, 150)
    x_all.append(x)
    i = 0
    for batch in datagen.flow(x, batch_size=1,
                          save_to_dir="E:/glass_class_dectect/data/无框眼镜1000/"):
       i += 1
       if i > 4:             # 数据扩充倍数，此处为数据扩充50倍
        break             #  否则生成器会退出循环