Feature matching result output

[Lucifer1002]

Hello, may I ask, how to output the matching results of image feature points?

[Shiaoming]

check this function:

Python-VO/utils/tools.py

Lines 38 to 91 in c03e88f

	def plot_matches(image0, image1, kpts0, kpts1, scores=None, layout="lr"):
	"""
	plot matches between two images. If score is nor None, then red: bad match, green: good match
	:param image0: reference image
	:param image1: current image
	:param kpts0: keypoints in reference image
	:param kpts1: keypoints in current image
	:param scores: matching score for each keypoint pair, range [0~1], 0: worst match, 1: best match
	:param layout: 'lr': left right; 'ud': up down
	:return:
	"""
	H0, W0 = image0.shape[0], image0.shape[1]
	H1, W1 = image1.shape[0], image1.shape[1]
	if layout == "lr":
	H, W = max(H0, H1), W0 + W1
	out = 255 * np.ones((H, W, 3), np.uint8)
	out[:H0, :W0, :] = image0
	out[:H1, W0:, :] = image1
	elif layout == "ud":
	H, W = H0 + H1, max(W0, W1)
	out = 255 * np.ones((H, W, 3), np.uint8)
	out[:H0, :W0, :] = image0
	out[H0:, :W1, :] = image1
	else:
	raise ValueError("The layout must be 'lr' or 'ud'!")
	kpts0, kpts1 = np.round(kpts0).astype(int), np.round(kpts1).astype(int)
	# get color
	if scores is not None:
	smin, smax = scores.min(), scores.max()
	assert (0 <= smin <= 1 and 0 <= smax <= 1)
	color = cm.gist_rainbow(scores * 0.4)
	color = (np.array(color[:, :3]) * 255).astype(int)[:, ::-1]
	else:
	color = np.zeros((kpts0.shape[0], 3), dtype=int)
	color[:, 1] = 255
	for (x0, y0), (x1, y1), c in zip(kpts0, kpts1, color):
	c = c.tolist()
	if layout == "lr":
	cv2.line(out, (x0, y0), (x1 + W0, y1), color=c, thickness=1, lineType=cv2.LINE_AA)
	# display line end-points as circles
	cv2.circle(out, (x0, y0), 2, c, -1, lineType=cv2.LINE_AA)
	cv2.circle(out, (x1 + W0, y1), 2, c, -1, lineType=cv2.LINE_AA)
	elif layout == "ud":
	cv2.line(out, (x0, y0), (x1, y1 + H0), color=c, thickness=1, lineType=cv2.LINE_AA)
	# display line end-points as circles
	cv2.circle(out, (x0, y0), 2, c, -1, lineType=cv2.LINE_AA)
	cv2.circle(out, (x1, y1 + H0), 2, c, -1, lineType=cv2.LINE_AA)
	return out

Originally posted by @Shiaoming in #7 (comment)