python opencv抠图 opencv 抠图自动 python

一.实现效果（效果不好 能用 ）：

二.py流程图

三.方法2. py代码

import cv2                #cv2库
import os
import tkinter.filedialog #tk 是ugi库 选择文件对话框的格式打开和保存图片库
import numpy as np        #NumPy系统是Python的一种开源的数值计算扩展
import matplotlib.pyplot as plt  #pyplot是常用的画图模块
from matplotlib import pyplot as plt
from PIL import Image     #图像处理模块

#文件对话框：
default_dir=r"文件路径"
 
path=tkinter.filedialog.askopenfilename(title=u'选择文件',initialdir=(os.path.expanduser((default_dir))))
#path='C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\car.jpg'
def show_pic(path): 
    #读入图
    initial_img = cv2.imread(path) #(600, 800, 3)  行，列，通道数 
    #initial_RGB = cv2.cvtColor(initial_img,cv2.COLOR_BGR2RGB)
    #灰度处理图像
    Gray_img = cv2.cvtColor(initial_img,cv2.COLOR_BGR2GRAY)
    #高斯去燥
    Gaussian = cv2.GaussianBlur(Gray_img, (5,5), 0, 0, cv2.BORDER_DEFAULT)
    #中值滤波
    Median = cv2.medianBlur(Gaussian, 5)
    #双边滤波
    #Test  = cv2.bilateralFilter(Gray_img,9,25,25)
    cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\111.jpg',Median)
    '''
    #sobel算子 轮廓
    # Sobel算子 XY方向求梯度
    x = cv2.Sobel(Median, cv2.CV_8U, 1, 0, ksize = 3) #X方向
    y = cv2.Sobel(Median, cv2.CV_8U, 0, 1, ksize = 3) #Y方向
    absX = cv2.convertScaleAbs(x)   # 转回uint8    
    absY = cv2.convertScaleAbs(y)    
    Sobel = cv2.addWeighted(absX, 0.5, absY, 0.5,0)
    '''
    #Canny边缘处理
    Canny = cv2.Canny(Median, 50, 200)
    #Otsu大津算法自适应阈值二值化
    ret, Binary = cv2.threshold(Canny,170, 255, cv2.THRESH_OTSU|cv2.THRESH_BINARY)
    
    #建立一个椭圆核函数,闭操作
    kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (25, 25))
    #执行图像形态学, 细节直接查文档，很简单
    closed = cv2.morphologyEx(Binary, cv2.MORPH_CLOSE, kernel)
    closed = cv2.erode(closed, None, iterations=4)
    closed = cv2.dilate(closed, None, iterations=4)
    #提取轮廓
    
    contours,_ = cv2.findContours(closed.copy(),cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)    
 
    #画出轮廓排序,cv2.contourArea 计算轮廓面积，将最大的那组轮廓挑出来
    c = sorted(contours, key=cv2.contourArea, reverse=True)[0]
    
    #compute the rotated bounding box of the largest contour
    rect = cv2.minAreaRect(c) #画出轮廓的最小外接矩形
 
        
    Box = np.int0(cv2.boxPoints(rect)) #获取最小外接矩形的4个顶点
        #按照Box画出轮廓    
    Final_img = cv2.drawContours(initial_img.copy(),[Box], -1, (255, 255, 255), -1)
    
    Xs = [i[0] for i in Box]
    Ys = [i[1] for i in Box]
    x1 = min(Xs)
    x2 = max(Xs)
    y1 = min(Ys)
    y2 = max(Ys)
    hight = y2 - y1
    width = x2 - x1
    crop_img = initial_img.copy()[y1:y1+hight, x1:x1+width]
#img2 =cv2.imread('C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\car.jpg')
#img_plate = path[y1:y2, x1:x2]

    #return crop_img
    
    img2 = cv2.imread('C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\car.jpg')    #获取原图给img2
# # 截取图像
    img_plate = img2[y1:y2, x1:x2]           
    cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\222.jpg', crop_img) #输出‘222.jpg’裁剪过的图片
    cv2.imshow("12",crop_img )
    cv2.waitKey(0)
    print("Crop_img1")
    print(crop_img)
    #cv2.imshow('Mask', path)
    #cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\7.jpg',path)
    #cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\7' + '.jpg', path)
    #cv.imwrite("C:\\Users\\Administrator\\Desktop\\demo\\py\\4\\7.jpeg",crop_img)

#print("Path2")
#print(path)
show_pic(path)

方法2测试图片：

四.方法1.1.py代码

import cv2
import numpy as np
import matplotlib.pyplot as plt
 
#1、imread加载图片
img = cv2.imread('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\image.jpg')
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\1yuantu.jpg', img)
 
#2、将图像转换为灰度图
 
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\2huitu.jpg', img)
#
#2、高斯平滑模糊
#GaussianBlur(InputArray img, OutputArray dst, Size ksize, double sigmaX, double sigmaY=0, int borderType=BORDER_DEFAULT )
#Size ksize必须为正奇数
img = cv2.GaussianBlur(img, (3, 3), 0, 0, cv2.BORDER_DEFAULT)
 
#3、中值滤波(池化)，消除噪音数据，medianBlur(InputArray src, OutputArray dst, int ksize)   ksize必须为奇数
img = cv2.medianBlur(img, 5)
 
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\3zhaoyin.jpg', img)
#4、利用Sobel方法可以进行sobel边缘检测，突出边缘
img = cv2.Sobel(img, cv2.CV_8U, 1, 0, ksize=3)
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\4bianyuan.jpg', img)
#图像的二值化就是将图像上的像素点的灰度值设置为0或255，这样将使整个图像呈现出明显的黑白效果，<150的全为黑，>150的全为白
ret, binary = cv2.threshold(img, 150, 255, cv2.THRESH_BINARY)
 
#膨胀，让轮廓突出
element1 = cv2.getStructuringElement(cv2.MORPH_RECT, (9, 7))
img = cv2.dilate(binary, element1, iterations=1)
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\5lunkuo.jpg', img)
#腐蚀
element2 = cv2.getStructuringElement(cv2.MORPH_RECT, (9, 1))
img = cv2.erode(img, element2, iterations=1)
#膨胀，让轮廓更明显
img = cv2.dilate(img, element1, iterations=3)
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\6pengzhang.jpg', img)
 
###############################################
# 查找轮廓(img: 原始图像，contours：矩形坐标点，hierarchy：图像层次)
contours, hierarchy = cv2.findContours(img, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\7pengzhang.jpg', img)
max_ratio = -1
ratios = []
num = 0
 
for i in range(len(contours)):
    cnt = contours[i]
 
    #计算轮廓面积
    area = cv2.contourArea(cnt)
    if area < 1000:
        continue
 
    #四边形的最小外接矩形,得到最小外接矩形的（中心(x,y), (宽,高), 旋转角度）
    rect = cv2.minAreaRect(cnt)
 
    # 矩形的四个坐标（顺序不定，但是一定是一个左下角、左上角、右上角、右下角这种循环顺序(开始是哪个点未知)）
    box = cv2.boxPoints(rect)
    # 转换为long类型
    box = np.int0(box)
 
    # 计算长宽高
    height = abs(box[0][1] - box[2][1])
    weight = abs(box[0][0] - box[2][0])
    ratio = float(weight) / float(height)
    # 正常的车牌宽高比在2.7~5之间
    if ratio > max_ratio:
        max_box = box
 
    if ratio > 5.5 or ratio < 2:
        continue
 
    num +=1
    ratios.append((max_box,ratio))
 
 
#返回就是车牌的矩阵的四个点的坐标
box = ratios[0][0]
#print(box)
#print(box[0,1])
 
ys = [box[0, 1], box[1, 1], box[2, 1], box[3, 1]]
#print(ys)
xs = [box[0, 0], box[1, 0], box[2, 0], box[3, 0]]
 
ys_sorted_index = np.argsort(ys)
#print(ys_sorted_index)
xs_sorted_index = np.argsort(xs)
 
# 获取x上的坐标
x1 = box[xs_sorted_index[0], 0]
#print(x1)
x2 = box[xs_sorted_index[3], 0]
#print(x2)
 
# 获取y上的坐标
y1 = box[ys_sorted_index[0], 1]
#print(y1)
y2 = box[ys_sorted_index[3], 1]
#print(y2)
#
img2 = cv2.imread('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\image.jpg')
# # 截取图像
img_plate = img2[y1:y2, x1:x2]
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\8.jpg', img_plate)
print(img_plate)
cv2.imshow("12",img_plate )
cv2.waitKey(0)

五.方法1.2.py代码

# -*- coding: utf-8 -*-
 
import cv2
import numpy as np
 
# 读取图片
imagePath = 'C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\image.jpg'
img = cv2.imread(imagePath)
 
# 转化成灰度图
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)


 
# 利用Sobel边缘检测生成二值图
sobel = cv2.Sobel(gray, cv2.CV_8U, 1, 0, ksize=3)
# 二值化
ret, binary = cv2.threshold(sobel, 0, 255, cv2.THRESH_OTSU + cv2.THRESH_BINARY)
 
# 膨胀、腐蚀
element1 = cv2.getStructuringElement(cv2.MORPH_RECT, (30, 9))
element2 = cv2.getStructuringElement(cv2.MORPH_RECT, (24, 6))
 
# 膨胀一次，让轮廓突出
dilation = cv2.dilate(binary, element2, iterations=1)
 
# 腐蚀一次，去掉细节
erosion = cv2.erode(dilation, element1, iterations=1)
 
# 再次膨胀，让轮廓明显一些
dilation2 = cv2.dilate(erosion, element2, iterations=2)
 
#  查找轮廓和筛选文字区域
region = []
contours, hierarchy = cv2.findContours(dilation2, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
for i in range(len(contours)):
    cnt = contours[i]
 
    # 计算轮廓面积，并筛选掉面积小的
    area = cv2.contourArea(cnt)
    if (area < 1000):
        continue
 
    # 找到最小的矩形
    rect = cv2.minAreaRect(cnt)
    print ("rect is: ")
    print (rect)
 
    # box是四个点的坐标
    box = cv2.boxPoints(rect)
    box = np.int0(box)
 
    # 计算高和宽
    height = abs(box[0][1] - box[2][1])
    width = abs(box[0][0] - box[2][0])
 
    # 根据文字特征，筛选那些太细的矩形，留下扁的
    if (height > width * 1.3):
        continue
 
    region.append(box)

cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\1.jpg', img)    
 
 
# 绘制轮廓
for box in region:
    cv2.drawContours(img, [box], 0, (0, 255, 0), 2)

cv2.imshow('img', img)
cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\2.jpg', img)
cv2.waitKey(0)
cv2.destroyAllWindows()   
    
#    box = region[0][0]
#print(box)
#print(box[0,1])
 
#    ys = [box[0, 1], box[1, 1], box[2, 1], box[3, 1]]
#print(ys)
 #   xs = [box[0, 0], box[1, 0], box[2, 0], box[3, 0]]
 
  #  ys_sorted_index = np.argsort(ys)
#print(ys_sorted_index)
   # xs_sorted_index = np.argsort(xs)

# 获取x上的坐标
    #x1 = box[xs_sorted_index[0], 0]
#print(x1)   
   # x2 = box[xs_sorted_index[3], 0]
#print(x2)
 
# 获取y上的坐标
    #y1 = box[ys_sorted_index[0], 1]
#print(y1)
    #y2 = box[ys_sorted_index[3], 1]
#print(y2)

    #img2 = cv2.imread('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\image.jpg')
# # 截取图像
    #img_plate =img2[y1:y2, x1:x2]
    #cv2.imwrite('C:\\Users\\Administrator\\Desktop\\demo\\py\\1\\10.jpg', img_plate)

#cv2.imshow('img', img)
#cv2.waitKey(0)
#cv2.destroyAllWindows()

方法1-1.2测试图片：

六.

图片有部分是手拍，有部分是网图，代码核心是看的别人的 时间很长了 忘了 对原作者说声抱歉。