如何在OpenCV中绘制抗锯齿圆形？Python实现问题求助

问题描述 #

尝试用Python的OpenCV在图片上绘制边缘平滑的圆形，但当前代码生成的圆形边缘不够精确。代码如下：

import cv2
import numpy as np
import imutils
from PIL import Image, ImageDraw, ImageFont  # 补充原代码缺失的导入

text1 = "10x"
text2 = "20gr"

# Load image in OpenCV
image = cv2.imread('Sasa.jfif')
resized = imutils.resize(image, width=500)
cv2.circle(resized,(350,150),65,(102,51,17),thickness=-1)

# Convert the image to RGB (OpenCV uses BGR)
cv2_im_rgb = cv2.cvtColor(resized,cv2.COLOR_BGR2RGB)

# Pass the image to PIL
pil_im = Image.fromarray(cv2_im_rgb)

draw = ImageDraw.Draw(pil_im)
# use a truetype font
font1 = ImageFont.truetype("arial.ttf", 50)
font2 = ImageFont.truetype("arial.ttf", 25)

# Draw the text
draw.text((310,110), text1, font=font1)
draw.text((325,170), text2, font=font2)

# Get back the image to OpenCV
cv2_im_processed = cv2.cvtColor(np.array(pil_im), cv2.COLOR_RGB2BGR)

cv2.imshow('Fonts', cv2_im_processed)
cv2.waitKey(1)

优化方案 #

1. 启用OpenCV抗锯齿绘制 #

OpenCV的cv2.circle支持通过lineType参数启用抗锯齿，直接修改绘制代码即可提升边缘平滑度：

# 加入lineType=cv2.LINE_AA启用抗锯齿
cv2.circle(resized, (350,150), 65, (102,51,17), thickness=-1, lineType=cv2.LINE_AA)

该参数对填充圆形（thickness=-1）同样有效，能让边缘像素过渡更自然，减少锯齿感。

2. 高分辨率画布绘制后缩放 #

如果需要更高精度的边缘，可以先在2-4倍分辨率的画布上绘制圆形，再缩小到目标尺寸，利用插值算法平滑边缘：

# 创建2倍分辨率的画布
scale = 2
high_res = cv2.resize(resized, (resized.shape[1]*scale, resized.shape[0]*scale), interpolation=cv2.INTER_LANCZOS4)
# 在高分辨率画布上画圆（坐标和半径按比例放大）
cv2.circle(high_res, (350*scale,150*scale), 65*scale, (102,51,17), thickness=-1, lineType=cv2.LINE_AA)
# 缩小回原尺寸
resized = cv2.resize(high_res, (resized.shape[1], resized.shape[0]), interpolation=cv2.INTER_LANCZOS4)

这种超采样方法能大幅提升边缘细节，适合对精度要求极高的场景。

3. 使用PIL/Pillow绘制圆形 #

既然已经在用PIL处理文字，也可以直接用PIL绘制圆形，ImageDraw.ellipse默认支持抗锯齿，边缘过渡更柔和：

# 跳过OpenCV的circle绘制，直接在PIL图像上画圆
draw = ImageDraw.Draw(pil_im)
center_x, center_y, radius = 350, 150, 65
# 注意PIL使用RGB颜色顺序，与OpenCV的BGR相反
draw.ellipse([center_x - radius, center_y - radius, center_x + radius, center_y + radius], fill=(17,51,102))

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