x = tf.reshape(x, shape=[-1, 28, 28, 1])的理解

Mnist中数据，输入n个样本，每个样本是784个列构成的向量。所以输入的是n*784的矩阵。但是输入到CNN中需要卷积，需要每个样本都是矩阵。

   x = tf.reshape(x, shape=[-1, 28, 28, 1])

newshape : int or tuple of ints
The new shape should be compatible with the original shape. If an integer, then the result will be a 1-D array of that length. One shape dimension can be -1. In this case, the value is inferred from the length of the array and remaining dimensions.
解释：如果等于-1的话，那么Numpy会根据剩下的维度计算出数组的另外一个shape属性值。

如何理解？举个栗子：

z = np.array([[1, 2, 3, 4],
	          [5, 6, 7, 8],
	          [9, 10, 11, 12],
	          [13, 14, 15, 16]])
Z = z.reshape([-1, 2, 2, 1])

最终输出结果：

[[[[ 1]
   [ 2]]
[[ 3]
   [ 4]]]
[[[ 5]
   [ 6]]
[[ 7]
   [ 8]]]
[[[ 9]
   [10]]
[[11]
   [12]]]
[[[13]
   [14]]
[[15]
   [16]]]]

这样做的目的是：将n个784个向量，变成n个28*28的
参考文章