import math
import cv2
import numpy as np
import matplotlib.pyplot as plt
plt.rcParams['font.sans-serif'] = ['FangSong']  # 設(shè)置字體以便正確顯示漢字
plt.rcParams['axes.unicode_minus'] = False  # 正確顯示連字符
MEDTH_ID=8
# 計(jì)算圖片清晰度
def getImageVar(img):
    img2gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)  # 轉(zhuǎn)化成灰度圖
    # 對(duì)圖片用 3x3 拉普拉斯算子做卷積得到邊緣  計(jì)算出方差，并最后返回。
    # 函數(shù)求完導(dǎo)數(shù)后會(huì)有負(fù)值，還有會(huì)大于255的值。而原圖像是uint8，即8位無(wú)符號(hào)數(shù)，所以建立的圖像位數(shù)不夠，會(huì)有截?cái)?。因此要使?4位有符號(hào)的數(shù)據(jù)類(lèi)型，即 cv2.CV_64F。
    # 再用var函數(shù)求方差
    imageVar = cv2.Laplacian(img2gray, cv2.CV_64F).var()
    return imageVar
# 轉(zhuǎn)接器
def handle(idx, img):
    if idx == 1: return handle_specific(img, linear) # 線(xiàn)性變化
    if idx == 2: return handle_specific(img, linear_up) # 分段線(xiàn)性變化
    if idx == 3: return handle_specific(img, Logarithmic) # 對(duì)數(shù)變換
    if idx == 4: return handle_specific(img, power) # 冪指變換
    if idx == 5: return handle_specific(img, cv2.equalizeHist)  # 直方圖均衡化
    if idx == 6: return handle_specific(img, auto_equalizeHist)  # 自適應(yīng)直方圖均衡化
    if idx == 7: return handle_specific(img, laplacian)  # laplacian算子圖像銳化
    if idx == 8: return handle_specific(img, non_sharpening)  # 非銳化掩蔽
# 處理函數(shù)
def handle_specific(img, func):
    img_list = [func(i) for i in cv2.split(img)]
    result = cv2.merge((img_list[0], img_list[1], img_list[2]))
    return result
# 線(xiàn)性變化
def linear(img):
    a, b = 1.5, 0
    for i in range(img.shape[0]):
        for j in range(img.shape[1]):
            if img[i][j] * a + b > 255:
                img[i][j] = 255
            else:
                img[i][j] = img[i][j] * a + b
    return img
# 分段線(xiàn)性變換-線(xiàn)性對(duì)比度拉伸,增強(qiáng)感興趣區(qū)域
def linear_up(img):
    # 灰度值的最大最小值
    r_min, r_max = 255, 0
    for i in range(img.shape[0]):
        for j in range(img.shape[1]):
            if img[i, j] > r_max:
                r_max = img[i, j]
            if img[i, j] < r_min:
                r_min = img[i, j]
    r1, s1 = r_min, 0
    r2, s2 = r_max, 255
    k = (s2 - s1) / (r2 - r1)
    for i in range(img.shape[0]):
        for j in range(img.shape[1]):
            if r1 <= img[i, j] <= r2:
                img[i, j] = k * (img[i, j] - r1)
    return img
# 對(duì)數(shù)變換
def Logarithmic(img):
    for i in range(img.shape[0]):
        for j in range(img.shape[1]):
            img[i][j] = math.log(1+img[i][j])
    cv2.normalize(img, img, 0, 255, cv2.NORM_MINMAX)
    img = cv2.convertScaleAbs(img)
    return img
# 對(duì)數(shù)變換
def power(img):
    for i in range(img.shape[0]):
        for j in range(img.shape[1]):
            img[i][j] = math.pow(img[i][j],1.2)
    cv2.normalize(img, img, 0, 255, cv2.NORM_MINMAX)
    img = cv2.convertScaleAbs(img)
    return img
# 自適應(yīng)的直方圖均衡化-非線(xiàn)性的對(duì)比度拉伸,增強(qiáng)感興趣區(qū)域
def auto_equalizeHist(img):
    clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
    img = clahe.apply(img)
    return img
def laplacian(img):
    kernel = np.array([[0, 1, 0], [1, -4, 1], [0, 1, 0]])  # laplacian卷積核的一個(gè)模板
    lapkernel_img = cv2.filter2D(img, -1, kernel) # 做卷積
    img = img - lapkernel_img
    return img
def non_sharpening(img):
    blur_img = cv2.blur(img, (5, 5))
    mask_img = img - blur_img
    img = img + mask_img
    return img
img = cv2.imread(filename='img/CB.61.20211203152034_crop_0.jpg', flags=1)
result = handle(MEDTH_ID, img)
print('原圖的清晰度：', getImageVar(img))
print('處理之后的清晰度', getImageVar(result))
fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(10, 8), dpi=100)
axes[0].imshow(img)
axes[0].set_title("原圖")
axes[1].imshow(result)
axes[1].set_title("處理之后的圖片")
plt.show()