numpy保存为图片方法及速度对比

numpy保存为图片方法及速度对比
最近遇到一个问题，有大量小图片需要保存，分析了一下代码，发现写入图片这个操作很耗时。于是想对比几种存储方法的性能。
参赛选手有：opencv、matplotlib、PIL、imageio

import numpy as np
import cv2
import matplotlib.pyplot as plt
from PIL import Image
import imageio
from line_profiler import LineProfiler

def saveimg(img):
    for i in range(100):
        #CV2
        cv2.imwrite('datatest/' + str(i) + '_cv2.jpg', img)
        #matplotlib
        plt.imsave('datatest/' + str(i) + '_matplotlib.jpg', img)
        #PIL
        im = Image.fromarray(img)
        im.save('datatest/' + str(i) + '_PIL.jpg')
        #imageio
        imageio.imsave('datatest/' + str(i) + '_imageio.jpg', img)



if __name__ == '__main__':
    img = np.random.randint(0, 255, size = [100,100,3], dtype ='uint8' )
    lp = LineProfiler()
    lp_wrapper = lp(saveimg)
    lp_wrapper(img)
    lp.print_stats()
    # saveimg(img)

输入结果看：

Line #      Hits         Time  Per Hit   % Time  Line Contents
==============================================================
     8                                           def saveimg(img):
     9       101       1756.0     17.4      0.0      for i in range(100):
    10                                                   #CV2
    11       100     604846.0   6048.5     14.2          cv2.imwrite('datatest/' + str(i) + '_cv2.jpg', img)    
    12                                                   #matplotlib
    13       100    1426342.0  14263.4     33.4          plt.imsave('datatest/' + str(i) + '_matplotlib.jpg', img)
    14                                                   #PIL
    15       100      70824.0    708.2      1.7          im = Image.fromarray(img)
    16       100     847999.0   8480.0     19.8          im.save('datatest/' + str(i) + '_PIL.jpg')
    17                                                   #imageio
    18       100    1320907.0  13209.1     30.9          imageio.imsave('datatest/' + str(i) + '_imageio.jpg', img)

opencv最快，14%
其次是PIL，21%
然后是imageio，30%
最慢的是plt，33%

和想象中的结果差不多。