咸鱼哥学python OpenCV之基本操作

秒速五厘米 2023-02-23 07:47 17阅读 0赞

使用的图片为lena.png

## 1、图像读取 ##

img_path = r'F:\Python_study\python_image\opencv\image\lena.png'
    noise_path = r'F:\Python_study\python_image\opencv\image\lenanoise.png'
    img = cv.imread(img_path)
    noise_img = cv.imread(noise_path)
    print(img.shape)
    cv.imshow('input', img)

# 2、获取感兴趣区域ROI #

img = cv.imread(img_path)
    roi_img = img[50:100,50:100]
    cv.imshow('roi-img', roi_img)

## 3、通道获取 ##

img = cv.imread(img_path)
    b,g,r = cv.split(img)
    print(b.shape, g.shape, r.shape)

## 4、通道合并 ##

img = cv.merge((b, g, r))

## 5、通道修改 ##

img = cv.imread(img_path)
    cur_img = img.copy()
    cur_img[:,:,1] = 0
    cur_img[:,:,2] = 0

## 6、边界填充 ##

top_size, bottom_size, left_size, right_size = (50,50,50,50)
    # 填充模式
    replicate = cv.copyMakeBorder(img, top_size, bottom_size, left_size, right_size, borderType=cv.BORDER_REFLECT)

填充模式(borderType)有以下几种：

## 7、数值计算 ##

cur2_img = img.copy()
    cur2_img  = cur2_img + 10
    print(cur2_img)
    print(cur2_img+cur_img)
    print(cv.add(cur_img, cur2_img))

## 8、图像融合 ##

# 首先图像的大小要保持一致才可以
    cur3_img = img.copy()
    cur3_img = cv.resize(cur3_img, (500,500))
    print(cur3_img.shape)
    res = cv.addWeighted(cur_img,0.4, cur2_img, 0.2,0)
    plt.imshow(res)

## 9、图像降噪 ##

blur_img = cv.blur(noise_img,(5,5))# 普通
    blur_img = cv.GaussianBlur(noise_img,(5,5), 1)# 高斯
    median_img = cv.medianBlur(noise_img, 5) # 均值
    # 使用矩阵运算进行组合输出
    # noise_blur_img = np.hstack((blur_img, blur_img, median_img))
    # cv.imshow('noise_blur_img',noise_blur_img)

## 10.1-图像腐蚀 ##

erosion_path = r'F:\Python_study\python_image\opencv\image\xianyuge.png'
    erosion_path_img = cv.imread(erosion_path)
    kernel = np.ones((2,2), np.uint8)
    erosion1 = cv.erode(erosion_path_img, kernel=kernel, iterations=1)
    erosion2 = cv.erode(erosion_path_img, kernel=kernel, iterations=2)
    erosion3 = cv.erode(erosion_path_img, kernel=kernel, iterations=3)
    # erosion_img = np.hstack((erosion1, erosion2, erosion3))
    # cv.imshow('erosion', erosion_img)

## 9.2-图像膨胀 ##

erosion_img_01 = cv.erode(erosion_path_img, kernel=kernel, iterations=1)
    dilate_img_01 = cv.dilate(erosion_img_01, kernel=kernel, iterations=1)
    dilate_img_02 = cv.dilate(erosion_img_01, kernel=kernel, iterations=2)
    dilate_img_03 = cv.dilate(erosion_img_01, kernel=kernel, iterations=3)
    # dilate_img = np.hstack((dilate_img_01, dilate_img_02, dilate_img_03))
    # # dilate_img = cv.erode(dilate_img, kernel=kernel, iterations=3)
    # # dilate_img = cv.dilate(dilate_img, kernel=kernel, iterations=2)
    # cv.imshow('dilate_img', dilate_img)

## 10、图像的形态形操作 ##

**10.1-开运算**

# 开运算 先腐蚀后膨胀
    path = r'F:\Python_study\python_image\opencv\image\xianyuge.png'
    img = cv.imread(path)
    kernel = np.ones((2,2), np.uint8)
    opening = cv.morphologyEx(img, cv.MORPH_OPEN, kernel=kernel)
    # cv.imshow('opening', opening)

**10.2-闭运算**

# 闭运算 先膨胀后腐蚀
    closing = cv.morphologyEx(img, cv.MORPH_CLOSE, kernel=kernel)
    # cv.imshow('closing', closing)

## 10.3梯度运算 ##

# 梯度=膨胀-腐蚀
    dilate = cv.dilate(img, kernel=kernel, iterations=2)
    erosion = cv.erode(img, kernel=kernel, iterations=2)
    res = np.hstack((dilate, erosion))
    # cv.imshow('res', res)
    
    
    # 梯度
    gradient = cv.morphologyEx(img, cv.MORPH_GRADIENT, kernel=kernel)
    # cv.imshow('gradient', gradient)

## 11、礼帽与黑帽 ##

**11.1-礼帽**

# 礼帽 = 原始输入-开运算结果
    tophat = cv.morphologyEx(img, cv.MORPH_TOPHAT, kernel=kernel)
    cv.imshow('topht', tophat)

**11.2-黑帽**

# 黑帽 = 闭运算(线膨胀后腐蚀) - 原始输入
    blactmat = cv.morphologyEx(img, cv.MORPH_BLACKHAT, kernel=kernel)
    cv.imshow('blacemat', blactmat)

图像输出使用如下两句：

cv.waitKey(0)
    cv.destroyAllWindows()