tensorflow之扩充自己的图片样本库

学习谷歌的tensorflow一直为没有足够的样本而苦恼。

最近学习到可以采用旋转倾斜角度的不同得到不同的样本库。

该方法可解决学习深度学习让而没有足够多的样本库的苦恼。

############################################################################################  
#!/usr/bin/python2.7  
# -*- coding: utf-8 -*-  
#Author  : zhaoqinghui  
#Date    : 2016.5.11  
#Function: add image  
##########################################################################################
import tensorflow as tf
import numpy as np
import math
import cv2
import sys
import os
from scipy import ndimage
import random

###########################################################################################
#设置自己的参数
###########################################################################################
training_index = './traini.txt'
newlabel_index ='./newlabel.txt'
classnum=36
maxImageNum=360
#############################################################################################

def read_traing_list():
	train_image_dir = []
	train_label_dir = []
	reader = open(training_index)
	while 1:
		line = reader.readline()
		tmp = line.split(" ")
		if not line:
			break
		train_image_dir.append(tmp[0])
		train_label_dir.append(tmp[1][0:-1])
	#print train_image_dir[1:maxImageNum]
	#print train_label_dir[1:maxImageNum]
	reader.close()
	return train_image_dir, train_label_dir

def distort_image():
	train_image_dir, train_label_dir = read_traing_list()
	label_reader = open(newlabel_index,"w")
	for idx in range(len(train_image_dir)):
		image_path = str(train_image_dir[idx])
		image_tmp = cv2.imread(image_path,cv2.IMREAD_GRAYSCALE)
		rotate_image = ndimage.rotate(image_tmp,random.randint(7,30))
		rotate_image = cv2.resize(rotate_image,(28,28))
		rotate_image_path = image_path[:-4]+"_1.png"
		print rotate_image_path
		cv2.imwrite(rotate_image_path,rotate_image)
		rotate_image2 = ndimage.rotate(image_tmp,random.randint(330,355))
		rotate_image2 = cv2.resize(rotate_image2,(28,28))
		rotate_image_path2 = image_path[:-4]+"_2.png"
		cv2.imwrite(rotate_image_path2,rotate_image2)
		label_reader.write(rotate_image_path+" "+str(train_label_dir[idx])+"\n")
		label_reader.write(rotate_image_path2+" "+str(train_label_dir[idx])+"\n")
	label_reader.close()
	print "done"
if __name__="__main__":
        distort_image()