交流群里的两个实例--直接放源码了-CFANZ编程社区

导读

本文主要介绍交流群里的两个实例，直接放源码。（公众号：OpenCV与AI深度学习）

实例一

要求：识别下图中加粗的文本内容。

交流群里的两个实例--直接放源码了_公众号

实现步骤：

【1】闭运算减少线条干扰

交流群里的两个实例--直接放源码了_公众号_02


import numpy as np
import cv2
img= cv2.imread('test.jpg')
#cv2.imshow('src', img)

gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
_,thres = cv2.threshold(gray,0,255,cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU)

k1=np.ones((9,9), np.uint8) 
close = cv2.morphologyEx(gray, cv2.MORPH_CLOSE, k1)
 
cv2.imwrite('close.jpg', close)
cv2.imshow('close', close)
cv2.waitKey(0)
cv2.destroyAllWindows()
print("Done!")

【2】OCR识别，参考下面以前的文章即可

让OCR更简单 | PaddleOCR+OpenCV实现文字识别步骤与代码演示

实战 | OpenCV+OCR实现环形文字识别实例(详细步骤 + 代码)

交流群里的两个实例--直接放源码了_深度学习_03

实例二

要求：识别下图中圆形锡点的数量和位置。

交流群里的两个实例--直接放源码了_灰度图_04

实现步骤：【1】转为灰度图、中值滤波

交流群里的两个实例--直接放源码了_深度学习_05

【2】灰度图和滤波图差分、阈值分割

交流群里的两个实例--直接放源码了_灰度图_06

交流群里的两个实例--直接放源码了_公众号_07

【3】形态学处理交流群里的两个实例--直接放源码了_公众号_08

【4】轮廓筛选、标注结果

交流群里的两个实例--直接放源码了_灰度图_09

完整源码：


import numpy as np
import cv2

font=cv2.FONT_HERSHEY_SIMPLEX

img = cv2.imread('AA.png')

cv2.imshow('src', img)

gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

blur = cv2.medianBlur(gray,15)

cv2.imwrite('blur.jpg',blur)

cv2.imshow('blur', blur)

diff = cv2.absdiff(gray, blur)

cv2.imshow('diff', diff)
cv2.imwrite('diff.jpg',diff)

ret,thres = cv2.threshold(diff,15,255,cv2.cv2.THRESH_BINARY)

cv2.imshow('thres', thres)
cv2.imwrite('thres.jpg',thres)

k1 = np.zeros((19, 19),np.uint8)
cv2.circle(k1,(9,9),9,(1,1,1),-1,cv2.LINE_AA)

closing = cv2.morphologyEx(thres, cv2.MORPH_CLOSE, k1, None, None, 1)#闭运算
cv2.imshow('closing',closing)

k2 = np.zeros((13, 13),np.uint8)
cv2.circle(k2,(6,6),6,(1,1,1),-1,cv2.LINE_AA)

opening = cv2.morphologyEx(closing, cv2.MORPH_OPEN, k2, None, None, 1)#闭运算
cv2.imshow('opening',opening)
cv2.imwrite('opening.jpg',opening)

contours,hierarchy = cv2.findContours(opening, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
count = 0
for cnt in contours:
  center,radius = cv2.minEnclosingCircle(cnt)
  if radius < 15 or radius > 30 or center[1] > 750 or (center[1] > 300 and center[1] < 520):
    continue
  count = count + 1
  ROI = opening[int(center[1]-(radius)):int(center[1]+radius),int(center[0]-radius):int(center[0]+radius)]
  dt = cv2.distanceTransform(ROI,cv2.DIST_L2,5,cv2.DIST_LABEL_PIXEL)
  transImg = cv2.convertScaleAbs(dt)
  cv2.normalize(transImg, transImg, 0, 255, cv2.NORM_MINMAX)
  _, _, _, max_loc = cv2.minMaxLoc(transImg)
  center = (max_loc[0]+center[0]-radius,max_loc[1]+center[1]-radius)
  #cv2.circle(img,(int(center[0]),int(center[1])),int(radius),(0,255,0),2)
  #cv2.circle(img,(int(center[0]),int(center[1])),24,(0,255,0),2)
  cv2.circle(img,(int(center[0]),int(center[1])),24,(0,255,0),2)
 
  cv2.drawMarker(img,(int(center[0]),int(center[1])),(0,0,255),
                     cv2.MARKER_CROSS, 20,2,8)
strCount = "count = %d" % count
cv2.putText(img,strCount,(10,60),font,1.5,(0,255,255),3)

cv2.imshow('result', img)
cv2.imwrite('result.jpg', img)

cv2.waitKey(0)
cv2.destroyAllWindows()