如何将BMP矢量化为单条开放路径而非闭合路径？

我正在开展一个简单的图像处理项目，需将二值图像转换为类似Adobe Illustrator中钢笔工具绘制的开放路径。当前遇到的问题是导出的SVG为闭合路径对象，而非开放路径。

已执行步骤 #

• 输入二值图像
• 使用Scikit反转二值图像
• 使用Scikit对反转图像进行骨架化处理
• 转换为8位图像
• 导出为JPG格式
• 使用pillow将JPG转为BMP
• 使用potrace将BMP转为SVG

程序1 - 二值化、反转、骨架化等处理 #

from skimage.morphology import skeletonize
from skimage.util import invert
from skimage import data
import matplotlib.pyplot as plt
import cv2
import numpy as np
import svgwrite

# Reading BINARY input & converting to greyscale
image = cv2.imread("C:\\Users\\jackc\\OneDrive\\Desktop\\medial_skeleton\\input2.JPG", cv2.IMREAD_COLOR)
greyscale = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)

# Threshold the grayscale image to get a binary image
_, binary_image = cv2.threshold(greyscale, 127, 255, cv2.THRESH_BINARY)

# Invert binary image
inversion = invert(binary_image)

# "Skeletonization"
skeleton = skeletonize(inversion)

# Convert image to 8-bit
skeleton_8bit = np.uint8(skeleton * 255)

# Increase line thickness by dilating the image
kernel = np.ones((3,3),np.uint8)
dilated_skeleton = cv2.dilate(skeleton_8bit, kernel, iterations = 1)

# Create a new SVG drawing
 dwg = svgwrite.Drawing("C:\\Users\\jackc\\OneDrive\\Desktop\\medial_skeleton\\svg_output1.svg", profile='full')

# Iterate over the pixels in the dilated skeleton image
for y in range(skeleton_8bit.shape[0]):
    for x in range(skeleton_8bit.shape[1]):
        # If the pixel is white, add a line to the SVG
        if skeleton_8bit[y, x] > 0:
            dwg.add(dwg.line((x, y), (x+1, y), stroke=svgwrite.rgb(0, 0, 0, '%')))

# Save the SVG
dwg.save()

# Writing the image to a file
cv2.imwrite("C:\\Users\\jackc\\OneDrive\\Desktop\\medial_skeleton\\skeleton_output1.JPG", skeleton_8bit)

程序2 - JPG转BMP #

from PIL import Image

# Open the JPG image
image = Image.open('C:\\Users\\jackc\\OneDrive\\Desktop\\medial_skeleton\\output7.JPG')

# Save the image as a BMP file
image.save('C:\\Users\\jackc\\OneDrive\\Desktop\\medial_skeleton\\bitmap_output1.bmp')

程序3 - BMP转SVG #

import subprocess

# Define the Potrace command as a list of strings
command = ["potrace", "-s", "-o", "C:\\Users\\jackc\\OneDrive\\Desktop\\medial_skeleton\\vectorized_output1.svg", "C:\\Users\\jackc\\OneDrive\\Desktop\\medial_skeleton\\bitmap_output1.bmp"]

# Run the command
subprocess.run(command, check=True)

问题细节 #

• 输入图像：一张手写风格的二值线条图
• 当前问题：导出的SVG并非由控制点构成的单条路径，而是由多个闭合路径组成的对象
• 预期结果：生成由单排控制点构成的单条开放路径，效果类似Adobe Illustrator“线条图”图像描摹功能处理二值图像的结果

解决方案更新 #

更新：已通过AutoTrace库的中心线功能实现需求，该库可从图像中提取中心线并生成符合预期的开放路径SVG

内容的提问来源于stack exchange，提问作者jackcbowers