/
ex_corner_detection.py
108 lines (93 loc) · 5.36 KB
/
ex_corner_detection.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
#!/usr/bin/env python
#The MIT License (MIT)
#Copyright (c) 2020 Massimiliano Patacchiola
#
#THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
#MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
#CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
#SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
#Tested on Ubuntu 18.04.3 LTS, OpenCV 4.1.2, Python 3
#Comparison of 4 different methods on corner detection.
#Parameters set such that all the methods will find around 500 keypoints in the video.
#You need a video named "video.mp4" in your script folder for running the code.
#Videos sourced from: https://www.pexels.com/videos
import cv2
import numpy as np
from operator import itemgetter
#print(cv2.__version__)
video_capture = cv2.VideoCapture("./video.mp4")
fourcc = cv2.VideoWriter_fourcc(*'XVID')
#out = cv2.VideoWriter("./original.avi", fourcc, 24.0, (3840,2160))
out_harris = cv2.VideoWriter("./harris.avi", fourcc, 24.0, (3840,2160))
out_shitomasi = cv2.VideoWriter("./shitomasi.avi", fourcc, 24.0, (3840,2160))
out_fast = cv2.VideoWriter("./fast.avi", fourcc, 24.0, (3840,2160))
out_orb = cv2.VideoWriter("./orb.avi", fourcc, 24.0, (3840,2160))
while(True):
ret, frame = video_capture.read()
if(frame is None): break #check for empty frames (en of video)
frame_gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
##Harris
mask_harris = cv2.cornerHarris(np.float32(frame_gray), blockSize=2, ksize=3, k=0.04) #2, 3, 0.04 // 2, 5, 0.07
#mask_harris = cv2.dilate(mask_harris, None)
cutout = np.sort(mask_harris.flatten())[-500] #sort from smaller to higher, then take index for cutout
corners = np.where(mask_harris > cutout)
corners = zip(corners[0], corners[1])
kp = list()
for i in corners: kp.append(cv2.KeyPoint(i[1], i[0], 20))
frame_harris = cv2.drawKeypoints(frame_gray, kp, None, [0, 0, 255],
flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
##Shi-Tomasi
#maxCorners: Maximum number of corners to return
#qualityLevel: Parameter characterizing the minimal accepted quality of image corners.
#minDistance: Minimum possible Euclidean distance between the returned corners.
#blockSize: Size of an average block for computing a derivative covariation matrix over each pixel neighborhood.
corners = cv2.goodFeaturesToTrack(frame_gray, maxCorners=500, qualityLevel=0.01, minDistance=10, blockSize=2)
corners = np.int0(corners)
kp = list()
for i in corners: kp.append(cv2.KeyPoint(i.ravel()[0], i.ravel()[1], 20))
frame_shitomasi = cv2.drawKeypoints(frame_gray, kp, None, [0, 0, 255],
flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
##FAST
#It is several times faster than other existing corner detectors.
#But it is not robust to high levels of noise. It is dependent on a threshold.
#threshold: a threshold over the point to keep
#nonmaxSuppression: wheter non-maximum suppression is to be applied or not
#Neighborhood (three flags) cv.FAST_FEATURE_DETECTOR_TYPE_5_8 / 7_12 / 9_16
#frame_fast = np.copy(frame)
#Here I choose the magic number 135 for the threshold so that it finds around 500 corners.
fast = cv2.FastFeatureDetector_create(threshold=165, nonmaxSuppression=True,
type=cv2.FAST_FEATURE_DETECTOR_TYPE_9_16) #default is TYPE_9_16
kp = fast.detect(frame_gray, None)
#print(len(kp)) #use this print to check how many keypoints are found by FAST
##Uncomment these two lines if you want to randomly pick 500 keypoints
#indices = np.random.choice(len(kp), 500, replace=False)
#kp = itemgetter(*indices.tolist())(kp)
for i in kp: i.size=20 #changing the diameter to make it coherent with the other methods
frame_fast = cv2.drawKeypoints(frame_gray, kp, None, [0, 0, 255],
flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
##ORB (Oriented FAST and Rotated BRIEF)
#First it use FAST to find keypoints, then apply Harris corner measure
#to find top N points among them.
#nFeatures: maximum number of features to be retained
#scoreType: whether Harris score or FAST score to rank the features (default: Harris)
orb = cv2.ORB_create(nfeatures=500)
kp = orb.detect(frame_gray, None)
kp, des = orb.compute(frame_gray, kp)
frame_orb = cv2.drawKeypoints(frame_gray, kp, None, [0, 0, 255],
flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)
#Writing in the output file
#out.write(frame)
out_harris.write(frame_harris)
out_shitomasi.write(frame_shitomasi)
out_fast.write(frame_fast)
out_orb.write(frame_orb)
#Showing the frame and waiting for the exit command
#cv2.imshow('Original', frame) #show on window
cv2.imshow('Harris', frame_harris) #show on window
cv2.imshow('Shi-Tomasi', frame_shitomasi) #show on window
cv2.imshow('FAST', frame_fast) #show on window
cv2.imshow('ORB', frame_orb) #show on window
if cv2.waitKey(1) & 0xFF == ord('q'): break #Exit when Q is pressed
#Release the camera
video_capture.release()
print("Bye...")