opencv图像条形码区域识别检测（python）

opencv图像条形码区域检测（python）

目录

opencv学习—图像条形码区域检测（python）

一、检测方法流程如下

二、各个流程步骤讲解

三、代码以及结果

鸣谢

---

一、检测方法流程如下

二、各个流程步骤讲解

包含二维码原图像

原图像灰度化

img=cv2.imread("C:\\Users\\Lijian\\Desktop\\test.jpg",cv2.IMREAD_GRAYSCALE) 

黑帽操作

kernel = np.ones((1, 3), np.uint8)
img = cv2.morphologyEx(img, cv2.MORPH_BLACKHAT, kernel, anchor=(1, 0))

阈值分割

thresh, img = cv2.threshold(img, 10, 255, cv2.THRESH_BINARY)

形态学运算调整区域

kernel = np.ones((1, 5), np.uint8)
img = cv2.morphologyEx(img, cv2.MORPH_DILATE, kernel, anchor=(2, 0), iterations=2) 
img = cv2.morphologyEx(img, cv2.MORPH_CLOSE, kernel, anchor=(2, 0), iterations=2)  
kernel = np.ones((21, 35), np.uint8)
img = cv2.morphologyEx(img, cv2.MORPH_OPEN, kernel, iterations=1)

区域检测算法（检测区域，过滤噪声，绘制区域）

contours, hierarchy = cv2.findContours(img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
unscale = 1.0 / scale
if contours != None:
    for contour in contours:
        if cv2.contourArea(contour) <= 2000:
            continue
        rect = cv2.minAreaRect(contour)
        rect = \
            ((int(rect[0][0] * unscale), int(rect[0][1] * unscale)), \
             (int(rect[1][0] * unscale), int(rect[1][1] * unscale)), \
             rect[2])
        box = np.int0(cv2.boxPoints(rect))
        cv2.drawContours(img_out, [box], 0, (0, 255, 0), thickness = 2) 

三、代码以及结果

import cv2
import matplotlib.pyplot as plt
import numpy as np
cv2.__version__
img=cv2.imread("C:\\Users\\Lijian\\Desktop\\test.jpg",cv2.IMREAD_GRAYSCALE)
img_out=cv2.imread("C:\\Users\\Lijian\\Desktop\\test.jpg")
cv2.imshow('img',img)
cv2.waitKey(0)
cv2.destroyWindow
scale=800/img.shape[1]
img=cv2.resize(img,(int(img.shape[1]*scale),int(img.shape[0]*scale)))
cv2.imshow('img1',img)
cv2.waitKey(0);
cv2.destroyWindow
#blackhat
kernel = np.ones((1, 3), np.uint8)
img = cv2.morphologyEx(img, cv2.MORPH_BLACKHAT, kernel, anchor=(1, 0))

#sogliatura
thresh, img = cv2.threshold(img, 10, 255, cv2.THRESH_BINARY)
#operazioni  morfologiche
kernel = np.ones((1, 5), np.uint8)
img = cv2.morphologyEx(img, cv2.MORPH_DILATE, kernel, anchor=(2, 0), iterations=2) #dilatazione
img = cv2.morphologyEx(img, cv2.MORPH_CLOSE, kernel, anchor=(2, 0), iterations=2)  #chiusura

kernel = np.ones((21, 35), np.uint8)
img = cv2.morphologyEx(img, cv2.MORPH_OPEN, kernel, iterations=1)
cv2.imshow('img1',img)
cv2.waitKey(0);
cv2.destroyWindow
#estrazione dei componenti connessi
contours, hierarchy = cv2.findContours(img, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)

unscale = 1.0 / scale
if contours != None:
    for contour in contours:
        
        # se l'area non è grande a sufficienza la salto 
        if cv2.contourArea(contour) <= 2000:
            continue
        
        #estraggo il rettangolo di area minima (in formato (centro_x, centro_y), (width, height), angolo)
        rect = cv2.minAreaRect(contour)
        #l'effetto della riscalatura iniziale deve essere eliminato dalle coordinate rilevate
        rect = \
            ((int(rect[0][0] * unscale), int(rect[0][1] * unscale)), \
             (int(rect[1][0] * unscale), int(rect[1][1] * unscale)), \
             rect[2])
        
        
        #disegno il tutto sull'immagine originale
        box = np.int0(cv2.boxPoints(rect))
        cv2.drawContours(img_out, [box], 0, (0, 255, 0), thickness = 2)
        
cv2.imshow('img1',img_out)
cv2.waitKey(0);
cv2.destroyWindow

原图

结果图像

鸣谢

基于OpenCV的条形码区域分割

本文地址：https://blog.csdn.net/weixin_44690935/article/details/109028599