python识别围棋定位棋盘位置

最近需要做一个围棋识别的项目，首先要将棋盘位置定位出来，效果图如下：

效果图

原图

中间处理效果

最终结果

思路分析

我们利用python opencv的相关函数进行操作实现，根据棋盘颜色的特征，寻找到相关特征，将棋盘区域抠出来。最好从原始图像中将棋盘位置截取出来。

源码：定位棋盘位置

from PIL import ImageGrab
import numpy as np
import cv2
from glob import glob

imglist = sorted(glob("screen/*.jpg"))
for i in imglist:
# while 1:
   img = cv2.imread(i)
   image = img.copy()
   w,h,c = img.shape
   img2 =  np.zeros((w,h,c), np.uint8)
   img3 =  np.zeros((w,h,c), np.uint8)
   # img = ImageGrab.grab() #bbox specifies specific region (bbox= x,y,width,height *starts top-left)

hsv=cv2.cvtColor(img,cv2.COLOR_BGR2HSV)
   lower = np.array([10,0,0])
   upper = np.array([40,255,255])
   mask = cv2.inRange(hsv,lower,upper)
   erodeim = cv2.erode(mask,None,iterations=2)  # 腐蚀
   dilateim = cv2.dilate(erodeim,None,iterations=2)

img = cv2.bitwise_and(img,img,mask=dilateim)
   frame = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
   ret, dst = cv2.threshold(frame, 100, 255, cv2.THRESH_BINARY)
   contours,hierarchy = cv2.findContours(dst, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)

cv2.imshow("0",img)
   i = 0
   maxarea = 0
   nextarea = 0
   maxint = 0
   for c in contours:
       if cv2.contourArea(c)>maxarea:
           maxarea = cv2.contourArea(c)
           maxint = i
       i+=1

#多边形拟合
   epsilon = 0.02*cv2.arcLength(contours[maxint],True)
   if epsilon<1:
       continue

#多边形拟合
   approx = cv2.approxPolyDP(contours[maxint],epsilon,True)
   [[x1,y1]] = approx[0]
   [[x2,y2]] = approx[2]

checkerboard = image[y1:y2,x1:x2]
   cv2.imshow("1",checkerboard)
   cv2.waitKey(1000)

cv2.destroyAllWindows()

带保存图像

from PIL import ImageGrab
import numpy as np
import cv2
from glob import glob
import os

imglist = sorted(glob("screen/*.jpg"))
a=0
for i in imglist:
# while 1:
   a=a+1
   img = cv2.imread(i)
   image = img.copy()
   w,h,c = img.shape
   img2 =  np.zeros((w,h,c), np.uint8)
   img3 =  np.zeros((w,h,c), np.uint8)
   # img = ImageGrab.grab() #bbox specifies specific region (bbox= x,y,width,height *starts top-left)

hsv=cv2.cvtColor(img,cv2.COLOR_BGR2HSV)
   lower = np.array([10,0,0])
   upper = np.array([40,255,255])
   mask = cv2.inRange(hsv,lower,upper)
   erodeim = cv2.erode(mask,None,iterations=2)  # 腐蚀
   dilateim = cv2.dilate(erodeim,None,iterations=2)

img = cv2.bitwise_and(img,img,mask=dilateim)
   frame = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
   ret, dst = cv2.threshold(frame, 100, 255, cv2.THRESH_BINARY)
   contours,hierarchy = cv2.findContours(dst, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)

# 保存图片的地址
   img_file_1 = "./temp"
   # 确认上述地址是否存在
   if not os.path.exists(img_file_1):
       os.mkdir(img_file_1)

cv2.imshow("0",img)
   cv2.imwrite(img_file_1 + "/" + 'temp_％d.jpg'％a, img)
   i = 0
   maxarea = 0
   nextarea = 0
   maxint = 0
   for c in contours:
       if cv2.contourArea(c)>maxarea:
           maxarea = cv2.contourArea(c)
           maxint = i
       i+=1

#多边形拟合
   epsilon = 0.02*cv2.arcLength(contours[maxint],True)
   if epsilon<1:
       continue

#多边形拟合
   approx = cv2.approxPolyDP(contours[maxint],epsilon,True)
   [[x1,y1]] = approx[0]
   [[x2,y2]] = approx[2]

checkerboard = image[y1:y2,x1:x2]
   cv2.imshow("1",checkerboard)
   cv2.waitKey(1000)
   # 保存图片的地址
   img_file_2 = "./checkerboard"
   # 确认上述地址是否存在
   if not os.path.exists(img_file_2):
       os.mkdir(img_file_2)
   cv2.imwrite(img_file_2 + "/" + 'checkerboard_％d.jpg'％a, checkerboard)
cv2.destroyAllWindows()

来源：https://blog.csdn.net/mao_hui_fei/article/details/118363347