因為MNIST資料是28*28的黑底白字影象，而且輸入時要將其拉直，也就是可以看成1*784的二維張量（張量的值在0~1之間），所以我們要對圖片進行預處理操作，是圖片能被網路識別。

以下是程式碼部分

import tensorflow as tf
import numpy as np
from PIL import Image
import backward as bw
import forward as fw

def restore(testPicArr):
    with tf.Graph().as_default() as g:
        x = tf.placeholder(tf.float32, [None, fw.INPUT_NODES])
        y_ 
 
= tf.placeholder(tf.float32, [None, fw.OUTPUT_NODES])
        y = fw.get_y(x, None)
        preValue = tf.arg_max(y, 1)
        
        ema = tf.train.ExponentialMovingAverage(bw.MOVING_ARVERAGE_DECAY)
        ema_restore = ema.variables_to_restore()
        saver = tf.train.Saver(ema_restore)
        
        with tf.Session() as sess:
            tf.logging.set_verbosity(tf.logging.WARN)
 
#降低警告等級
            ckpt = tf.train.get_checkpoint_state("./model/")
            if ckpt and ckpt.model_checkpoint_path:
                saver.restore(sess, ckpt.model_checkpoint_path)
                
                preValue = sess.run(preValue, feed_dict = {x: testPicArr})
                return preValue
            
 
else:
                print("NO!!!")
                return -1
    
def pre_pic(picName):
    img = Image.open(picName)
    reIm = img.resize((28, 28), Image.ANTIALIAS)
    im_arr = np.array(reIm.convert('L'))#變為灰度圖
    threshold = 50#閾值，將圖片二值化操作
    for i in range(28):
        for j in range(28):
            im_arr[i][j] = 255 - im_arr[i][j]#進行反色處理
            if(im_arr[i][j] < threshold):
                im_arr[i][j] = 0
            else: im_arr[i][j] = 255
    
    nm_arr = im_arr.reshape([1,784])
    nm_arr = nm_arr.astype(np.float32)#型別轉換
    img_ready = np.multiply(nm_arr, 1.0/255.0)#把值變為0~1之間的數值
    
    return img_ready

def app():
    testNum = input("Input the number of test pictutre:")
    for i in range(int(testNum)):
        testPic = input("the path of test picture:")
        testPicArr = pre_pic(testPic)
        preValue = restore(testPicArr)
        print("The prediction number is :" , preValue)
        
def main():
    app()
    
if __name__ == '__main__':
    main()