Corner Detection
Summary
Corner Detection by Shi-Tomase with open cv examples.
Example 1. - detect corners on flat chessboard.
Example 2. - detect roi of text in an image.
import cv2
import numpy as np
import matplotlib.pyplot as plt
flat_chess = cv2.imread(file_path)
flat_chess = cv2.cvtColor(flat_chess,cv2.COLOR_BGR2RGB)
plt.figure(1)
plt.imshow(flat_chess)
flat_chess_gray = cv2.cvtColor(flat_chess,cv2.COLOR_RGB2GRAY)
corners =cv2.goodFeaturesToTrack(image=flat_chess_gray,
maxCorners=64,
qualityLevel=0.01,
minDistance=10)
flat_chess_new = flat_chess.copy()
corners = np.int0(corners)
for i in corners:
x,y = i.ravel()
cv2.circle(flat_chess_new,(x,y),3,255,-1)
plt.figure(2)
plt.imshow(flat_chess_new)
img = cv2.imread(file_path)
img = cv2.cvtColor(img,cv2.COLOR_BGR2RGB)
plt.figure(3)
plt.imshow(img)
img_gray = cv2.cvtColor(img,cv2.COLOR_RGB2GRAY)
corners =cv2.goodFeaturesToTrack(image=img_gray,
maxCorners=200,
qualityLevel=0.01,
minDistance=10)
corners = np.int0(corners)
x_lst,y_lst =[],[]
img_all_corners = img.copy()
for i in corners:
x,y = i.ravel()
x_lst.append(x)
y_lst.append(y)
cv2.circle(img_all_corners,( x,y ),3,255,-1)
plt.figure(4)
plt.imshow(img_all_corners)
values, counts = np.unique(corners[:,0,1], return_counts=True)
values_y = values[counts>=sorted(counts)[-4]]
x_lst1,y_lst1 =[],[]
img_selected = img.copy()
for idx,val in enumerate(y_lst):
if val in values_y:
x_lst1.append(x_lst[idx])
y_lst1.append(y_lst[idx])
cv2.circle(img_selected,(x_lst[idx],y_lst[idx]),3,255,-1)
plt.figure(5)
plt.imshow(img_selected)
y_min,y_max = min(y_lst1),max(y_lst1)
x_min,x_max = min(x_lst1),max(x_lst1)
space = 5
img_new = img.copy()
cv2.rectangle(img_new,
pt1=(x_min-space,y_min-space),
pt2=(x_max+space,y_max+space),
color=(255,0,0),
thickness=2)
plt.figure(6)
plt.imshow(img_new)
