Question

我发现PIL中有ImageOps.autocontrast的功能，在OpenCV中它对应的功能是什么？

编辑：

我按照pil的文档，以cv2和numpy来实现：

import PIL
from PIL import ImageOps, Image
import cv2
import numpy as np

def autocontrast(im, cutoff=0):
    channels = []
    for i in range(3):
        n_bins = 256
        hist = cv2.calcHist([im], [i], None, [n_bins], [0, n_bins])
        n = np.sum(hist)
        cut = cutoff * n // 100
        low = np.argwhere(np.cumsum(hist) > cut)
        low = low[0]
        high = np.argwhere(np.cumsum(hist[::-1]) > cut)
        high = n_bins - 1 - high[0]
        if high <= low:
            table = np.arange(n_bins)
        else:
            scale = (n_bins - 1) / (high - low)
            offset = -low * scale
            table = np.arange(n_bins) * scale + offset
            table[table < 0] = 0
            table[table > n_bins - 1] = n_bins - 1
        table = table.astype(np.uint8)
        channels.append(table[im[:, :, i]])
    out = cv2.merge(channels)
    return out

impth = './1_070323.jpg'
im = Image.open(impth)
imcv = cv2.imread(impth)
outcv = autocontrast(imcv, cutoff=10)[:, :, ::-1]
outpil = ImageOps.autocontrast(im, cutoff=10)
print(np.sum(outcv - np.array(outpil)))

现在，我测试了cv2 / numpy和pil方法的速度，发现cv2方法甚至比pil方法还要慢：

impth = './1_070323.jpg'
im = Image.open(impth)
imcv = cv2.imread(impth)
n_test = 100
t1 = time.time()
for i in range(n_test):
    outpil = ImageOps.autocontrast(im, cutoff=10)
t2 = time.time()
for i in range(n_test):
    outcv = autocontrast(imcv, cutoff=10)
t3 = time.time()

print('pil time: {}'.format(t2 - t1))
print('cv2 time: {}'.format(t3 - t2))

结果是pil需要大约0.8秒，而cv2需要3秒才能运行100次。我已经阅读了pil实现的源代码，其中充满了python for循环。怎么会比cv2 / numpy的实现更快呢？

PIL中自动对比度的对应OpenCV功能是什么？

0 个答案: