http://takatakamanbou.hatenablog.com/entry/2015/01/17/221234

mnist0117.py

import struct
import numpy as np


class MNIST:

    def __init__( self, LT ):

        if LT == 'L':
            self.fnLabel = 'train-labels-idx1-ubyte'
            self.fnImage = 'train-images-idx3-ubyte'
        else:
            self.fnLabel = 't10k-labels-idx1-ubyte'
            self.fnImage = 't10k-images-idx3-ubyte'

    def getLabel( self ):
        
        return _readLabel( self.fnLabel )

    def getImage( self ):
        
        return _readImage( self.fnImage )


##### reading the label file
#
def _readLabel( fnLabel ):

    f = open( fnLabel, 'r' )

    ### header (two 4B integers, magic number(2049) & number of items)
    #
    header = f.read( 8 )
    mn, num = struct.unpack( '>2i', header )  # MSB first (bigendian)
    assert mn == 2049
    #print mn, num

    ### labels (unsigned byte)
    #
    label = np.array( struct.unpack( '>%dB' % num, f.read() ), dtype = int )

    f.close()

    return label


##### reading the image file
#
def _readImage( fnImage ):

    f = open( fnImage, 'r' )

    ### header (four 4B integers, magic number(2051), #images, #rows, and #cols
    #
    header = f.read( 16 )
    mn, num, nrow, ncol = struct.unpack( '>4i', header ) # MSB first (bigendian)
    assert mn == 2051
    #print mn, num, nrow, ncol

    ### pixels (unsigned byte)
    #
    pixel = np.empty( ( num, nrow, ncol ) )
    npixel = nrow * ncol
    for i in range( num ):
        buf = struct.unpack( '>%dB' % npixel, f.read( npixel ) )
        pixel[i, :, :] = np.asarray( buf ).reshape( ( nrow, ncol ) )

    f.close()

    return pixel



if __name__ == '__main__':

    print '# MNIST training data'
    mnist = MNIST( 'L' )
    lab = mnist.getLabel()
    dat = mnist.getImage()
    print lab.shape, dat.shape

    print '# MNIST test data'
    mnist = MNIST( 'T' )
    lab = mnist.getLabel()
    dat = mnist.getImage()
    print lab.shape, dat.shape
    

mnistimage0117.py

import cv2
import numpy as np
import mnist0117 as mnist

nx = 10
ny = 10
gap = 4

mn = mnist.MNIST( 'L' )
dat = mn.getImage()[:nx*ny]
nrow, ncol = dat.shape[1:]

width  = nx * ( ncol + gap ) + gap
height = ny * ( nrow + gap ) + gap
img = np.zeros( ( height, width ), dtype = int ) + 128

for iy in range( ny ):
    lty = iy * ( nrow + gap ) + gap
    for ix in range( nx ):
        ltx = ix * ( ncol + gap ) + gap
        img[lty:lty+nrow, ltx:ltx+ncol] = dat[iy*nx+ix]

cv2.imwrite( 'hoge.png', img )