Below is a block-matching given for motion estimation.
A code for block-matching is given under

Understand this code and answer why optical flow is very useful and powerful, yet has some drawbacks! describe the main aspects based in the code given. Only rely to theory given in the lectures!

cap = cv2.VideoCapture(video_path)

suc, prev = cap.read()

prevgray = cv2.cvtColor(prev, cv2.COLOR_BGR2GRAY)

count=0

while count<=20:

suc, img = cap.read()

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

# start time to calculate FPS

start = time.time()

flow = cv2.calcOpticalFlowFarneback(prevgray, gray, None, 0.5, 3, 15, 3, 5, 1.2, 0)

prevgray = gray

# End time

end = time.time()

# calculate the FPS for current frame detection

fps = 1 / (end-start)

print(f"{fps:.2f} FPS")

cv2_imshow(motion_flow(gray, flow))

cv2_imshow(motion_hsv(flow))

key = cv2.waitKey(5)

if key == ord('q'):

break

count+=1
 

Transcribed Image Text:

Search Area Candidate Block Search Area Candidate Block