python - 使用 Python OpenCV 缩小和放大轮廓图像

Question

我有一个带有如下对象的图像：

我可以检测轮廓并得到一个只有球区域的蒙版，但我的 ROI 是边缘区域，这意味着我需要一个更大和更小的蒙版，它们结合起来得到这个：

所以我的问题是：如何缩小/放大轮廓中心周围的轮廓蒙版？

标签： pythonnumpyopencvcontouredge-detection

Answer 1

这是在 Python/OpenCV 中执行此操作的一种方法。

 - Read the input
 - Convert to grayscale
 - Threshold
 - Use morphology close and open to clean up noise and small regions to form a mask
 - Dilate the mask
 - Erode the mask
 - Merge the input and dilated mask
 - Merge the eroded mask with the previous result
 - Save the result

输入：

import cv2
import numpy as np

# read image
img = cv2.imread("basketball.png")

# convert img to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# make anything not white into black
mask = gray.copy()
mask[mask!=255] = 0

# invert mask so center is white and outside is black
mask = 255 - mask

# close open mask to clean up small regions and make 3 channels
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5,5))
mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, kernel)
mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE, kernel)
mask = cv2.cvtColor(mask, cv2.COLOR_GRAY2BGR)

# erode mask
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (51,51))
erode = cv2.morphologyEx(mask, cv2.MORPH_ERODE, kernel)

# dilate mask and make 3 channels
kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (51,51))
dilate = cv2.morphologyEx(mask, cv2.MORPH_DILATE, kernel)

# merge image onto dilated mask using mask
result = np.where(mask==(255,255,255), img, dilate)

# merge inverted erode onto result using erode
result = np.where(erode==(255,255,255), (255-erode), result)

# write result to disk
cv2.imwrite("basketball_mask.png", mask)
cv2.imwrite("basketball_eroded_mask.png", erode)
cv2.imwrite("basketball_dilate_mask.png", dilate)
cv2.imwrite("basketball_dilate_mask.png", dilate)
cv2.imwrite("basketball_result.png", result)

# display it
cv2.imshow("image", img)
cv2.imshow("mask", mask)
cv2.imshow("erode", erode)
cv2.imshow("dilate", dilate)
cv2.imshow("result", result)
cv2.waitKey(0)

面具：

侵蚀面具：

扩张面膜：

结果：

注意：如果你扩张太多，你会到达图像的边缘，然后形状会发生变化。为避免这种情况，请使用足够的背景颜色填充输入以包含扩张的大小。