传统的图像分割方法包含代码/注释/结果

import cv2 import numpy as np import matplotlib.pyplot as plt from scipy import ndimage as ndi from skimage import morphology, feature, color, data, filters #-----------------------------------------------------------# # 基于距离变换的分水岭分割 # 特适用两个目标连载一块 / 将其分开的情形 #-----------------------------------------------------------# def distance_water(image_path): #-------------------------------------------------------# # 读入图像 / 二值化 / 显示图像 #-------------------------------------------------------# # image = cv2.imread(image_path) # image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) # ret, image = cv2.threshold(image, 200, 255, cv2.THRESH_BINARY_INV) # # ret, image = cv2.threshold(image, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU) # cv2.namedWindow('image', 0) # cv2.imshow('image', image) # cv2.waitKey(0) #-------------------------------------------------------# # 想要更明显的看出效果使用下面的自己创建的两个挨在一起的圆 # 创建两个带有重叠圆的图像 #-------------------------------------------------------# x, y = np.indices((80, 80)) x1, y1, x2, y2 = 28, 28, 44, 52 r1, r2 = 16, 20 mask_circle1 = (x - x1) ** 2 + (y - y1) ** 2 < r1 ** 2 mask_circle2 = (x - x2) ** 2 + (y - y2) ** 2 < r2 ** 2 image = np.logical_or(mask_circle1, mask_circle2) #-------------------------------------------------------# # scipy 距离变化 / distance_transform_edt # 计算图像中非零点到最近背景点（即0）的距离 # peak_local_max函数返回图像中局部峰值的坐标 #-------------------------------------------------------# distance = ndi.distance_transform_edt(image) # print(distance[40:80, 40:80]) # print(distance[distance == 20].sum()) #-------------------------------------------------------# # peak_local_max返回和图像相同形状的数组， # 对其中的峰值点进行标记为True / 其余False #-------------------------------------------------------# local_maxi = feature.peak_local_max(distance, indices=False, footprint=np.ones((3, 3)), labels=image) # 寻找峰值 #-------------------------------------------------------# # ndi.label 返回 列表 / 峰值个数 markers = ndi.label(local_maxi)[0] # 初始标记点 # print(ndi.label(local_maxi)[0]) #-------------------------------------------------------# # 基于距离变换的分水岭算法 # 1. 以原始图像的前景区域为边界进行填充， # 2. 输入地势条件-distance，数值越低表示地势越低， # 3. markers为注水点标记， # 4. 注水时会以注水点标记为数值，以相应的类别值为水值进行注水。 #-------------------------------------------------------# labels = morphology.watershed(-distance, markers, mask=image) #-------------------------------------------------------# # 画图 / 显示 #-------------------------------------------------------# fig, axes = plt.subplots(nrows=2, ncols=2, figsize=(8, 8)) # 拉伸成一维数组 axes = axes.ravel() ax0, ax1, ax2, ax3 = axes ax0.imshow(image, cmap=plt.cm.gray, interpolation='nearest') ax0.set_title("Original") ax1.imshow(-distance, cmap=plt.cm.jet, interpolation='nearest') ax1.set_title("Distance") ax2.imshow(markers, cmap=plt.cm.winter, interpolation='nearest') ax2.set_title("Markers") ax3.imshow(labels, cmap=plt.cm.winter, interpolation='nearest') ax3.set_title("Segmented") for ax in axes: ax.axis('off') fig.tight_layout() plt.savefig('./image/distance_water.png') plt.show() #-----------------------------------------------------------# # 基于梯度的分水岭分割 # 使用于边缘较为明显，差异交大的图像分割 # 原理：1. 梯度图像在边缘处有较高的像素值，在其它地方则有较低的像素值， # 2. 理想情况下，分山岭恰好在边缘，因此可以根据梯度来寻找分山岭。 #-----------------------------------------------------------# def gradient_water(image_path): # image = cv2.imread(image_path) # image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) # ret, image = cv2.threshold(image, 200, 255, cv2.THRESH_BINARY_INV) # ret, image = cv2.threshold(image, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU) # cv2.namedWindow('image', 0) # cv2.imshow('image', image) # cv2.waitKey(0) #-------------------------------------------------------# # 1."data/camera.png" # 2. 过滤噪声 / morphology.disk()滤波器 # 3. 中值滤波 image = color.rgb2gray(data.camera()) denoised = filters.rank.median(image, morphology.disk(2)) # 将梯度值低于10的作为开始标记点 markers = filters.rank.gradient(denoised, morphology.disk(5)) < 10 markers = ndi.label(markers)[0] # 计算梯度 gradient = filters.rank.gradient(denoised, morphology.disk(2)) # 基于梯度的分水岭算法 labels = morphology.watershed(gradient, markers, mask=image) fig, axes = plt.subplots(nrows=2, ncols=2, figsize=(6, 6)) axes = axes.ravel() ax0, ax1, ax2, ax3 = axes ax0.imshow(image, cmap=plt.cm.gray, interpolation='nearest') ax0.set_title("Original") ax1.imshow(gradient, cmap=plt.cm.winter, interpolation='nearest') ax1.set_title("Gradient") ax2.imshow(markers, cmap=plt.cm.jet, interpolation='nearest') ax2.set_title("Markers") ax3.imshow(labels, cmap=plt.cm.jet, interpolation='nearest') ax3.set_title("Segmented") for ax in axes: ax.axis('off') fig.tight_layout() plt.savefig('./image/gradient_water.png') plt.show() if __name__ == "__main__": image_path = './image/1.JPG' # distance_water(image_path) gradient_water(image_path)