OpenCV图像处理技术之直方图处理

 Fu Xianjun. All Rights Reserved.所有素材来自于小傅老师。

任务一：直方图绘制

使用plt.hist绘制直方图

cv2.calcHist绘制直方图

任务二：使用掩膜进行直方图绘制

任务四：直方图均衡化

理解直图均衡化并能够利用函数使用函数calcHist实现。

任务五：直方图比较

直方图比较

创建 RGB 三通道直方图（直方图矩阵）

创建一个（16*16*16,1）的初始矩阵，作为直方图矩阵

 16*16*16的意思为三通道每通道有16个bins

人为构建直方图矩阵的索引，该索引是通过每一个像素点的三通道值进行构建

该处形成的矩阵即为直方图矩阵

创建第一幅图的rgb三通道直方图（直方图矩阵）

直方图比较函数

import cv2
import numpy as np
def create_rgb_hist(image):
    h, w, c = image.shape
    rgbhist = np.zeros([16 * 16 * 16, 1], np.float32)
    bsize = 256 / 16
    for row in range(h):
        for col in range(w):
            b = image[row, col, 0]
            g = image[row, col, 1]
            r = image[row, col, 2]
            index = int(b / bsize) * 16 * 16 + int(g / bsize) * 16 + int(r / bsize)
            rgbhist[int(index), 0] += 1
    return rgbhist

def hist_compare(image1, image2):
    hist1 = create_rgb_hist(image1)
    hist2 = create_rgb_hist(image2)
    match1 = cv2.compareHist(hist1, hist2, cv2.HISTCMP_BHATTACHARYYA)
    match2 = cv2.compareHist(hist1, hist2, cv2.HISTCMP_CORREL)
    match3 = cv2.compareHist(hist1, hist2, cv2.HISTCMP_CHISQR)
    print("巴氏距离：%s, 相关性：%s, 卡方：%s" %(match1, match2, match3))

img1 = cv2.imread("dog1.jpg")
cv2.imshow("dog1", img1)
img2 = cv2.imread("dog3.jpg")
cv2.imshow("dog3", img2)
hist_compare(img1, img2)
cv2.waitKey(0)
cv2.destroyAllWindows()

直方图阈值法

import cv2 
import numpy as np

#计算灰度直方图
def calcGrayHist(grayimage):
    #灰度图像矩阵的高，宽
    rows, cols = grayimage.shape
    print(grayimage.shape)
    #存储灰度直方图
    grayHist = np.zeros([256],np.uint64)
    for r in range(rows):
        for c in range(cols):
            grayHist[grayimage[r][c]] += 1
    return grayHist

#阈值分割：直方图阈值法 
def threshTwoPeaks(image):
    if len(image.shape) == 2:
        gray = image
    else:
        gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
        print(666666)
    #计算灰度直方图
    histogram = calcGrayHist(gray)
    #寻找灰度直方图的最大峰值对应的灰度值
    maxLoc = np.where(histogram==np.max(histogram))
    firstPeak = maxLoc[0][0]
    #寻找灰度直方图的第二个峰值对应的灰度值
    measureDists = np.zeros([256],np.float32)
    for k in range(256):
        measureDists[k] = pow(k-firstPeak,2)*histogram[k]
    maxLoc2 = np.where(measureDists==np.max(measureDists))
    secondPeak = maxLoc2[0][0]

    #找到两个峰值之间的最小值对应的灰度值，作为阈值
    thresh = 0
    if firstPeak > secondPeak:#第一个峰值再第二个峰值的右侧
        temp = histogram[int(secondPeak):int(firstPeak)]
        minloc = np.where(temp == np.min(temp))
        thresh = secondPeak + minloc[0][0] + 1
    else:#第一个峰值再第二个峰值的左侧
        temp = histogram[int(firstPeak):int(secondPeak)]
        minloc = np.where(temp == np.min(temp))
        thresh =firstPeak + minloc[0][0] + 1

    #找到阈值之后进行阈值处理，得到二值图
    threshImage_out = gray.copy()
    #大于阈值的都设置为255
    threshImage_out[threshImage_out > thresh] = 255
    threshImage_out[threshImage_out <= thresh] = 0
    return thresh, threshImage_out

if __name__ == "__main__":
    img = cv2.imread('dog2.png')
    thresh,threshImage_out = threshTwoPeaks(img)
    print(thresh)
    cv2.imshow('threshImage_out',threshImage_out) 
    cv2.waitKey(0)
    cv2.destroyAllWindows()

好了今天的学习就到这了吧

我们下期再见！