如何使用OpenCV在空白图像上绘制形状,使其与源图像完全相同?

问题描述

我有一个源图像(黑色背景和白色物体),我想从该图像中找到轮廓,创建一个新的具有黑色背景的空白图像,并像在源图像中一样一幅画轮廓图片。我知道如何查找轮廓并创建空白图像,但是在以与原始图像中相同的方式绘制轮廓时遇到了麻烦。我得到了一些结果,但是颜色填充有时填充不正确或看起来不正确。我一直在思考是否应该得到轮廓的倒数并填充它,但是我不确定如何做到这一点或是否有可能。我基本上想要做的是在视频上一张一张地绘制图像的形状。

这是我的代码

import cv2
import numpy as np
import sys
import statistics as st


# Function to denoise images
def denoise(contours):
    # Array that contains contours that will be removed
    filtered_cont = []
    sample = []
    # Gathering sample
    for contour in contours:
        sample.append(len(contour))
    # Calculating standard deviation
    stdev = st.stdev(sample)
    for contour in contours:
        # Comparing contour length to standard deviation divided by 3
        if len(contour) < stdev/3:
            # Appending small contours to an array
            filtered_cont.append(contour)
    return filtered_cont

img = cv2.imread(sys.argv[1])

# Getting image dimensions
h,w,l = img.shape
size = (w,h)

gray_image=cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
contours,hierarchy=cv2.findContours(gray_image,cv2.RETR_TREE,cv2.CHAIN_APPROX_NONE)

noisy_contours = denoise(contours)
# Filtering out noise
cv2.drawContours(img,noisy_contours,-1,color=(0,0),thickness=cv2.FILLED)
# Saving denoised image
cv2.imwrite('contours-denoised.png',img)


filtered=cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
edged=cv2.Canny(filtered,20,100)

# Creating blank black image
blank_image= np.zeros((h,3),np.uint8)

# Finding contours from denoised image
contours,hierarchy=cv2.findContours(edged,cv2.CHAIN_APPROX_NONE)
# Sorting contours from large to small
contours.sort(key=len,reverse=True)

# Initializing video output
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter('project.mp4',fourcc,10,size)

# Iterating through contours
for  cont in contours:
    # Drawing one contour and filling it with white
    cv2.drawContours(blank_image,[cont],color=(255,255,255),thickness=cv2.FILLED)
    # Writing image into video stream
    out.write(blank_image)
out.release()

# Saving image
cv2.imwrite('contours-denoised-drawn.png',blank_image)

Original image

源图像

Original image denoised

源图像去噪

Denoised image after Canny edge detection

坎尼边缘

Drawn version of denoised image

去噪图像的绘制版本(在Canny之后)

解决方法

暂无找到可以解决该程序问题的有效方法,小编努力寻找整理中!

如果你已经找到好的解决方法,欢迎将解决方案带上本链接一起发送给小编。

小编邮箱:dio#foxmail.com (将#修改为@)