Zoom Logo

MMAT5390202021 - Shared screen with speaker view
Ka Lung CHAN
19:59
YES
JIN Xg
20:00
yes
Chan Chan River
20:01
yes
James Tung
20:02
yes
Ka Lung CHAN
20:05
YES
Zilei HE
20:07
yes
Ho Renee TIN
20:14
Badaudio
Ho Renee TIN
20:18
only me?
Ka Lung CHAN
20:28
seems good for me
Ho Renee TIN
20:40
I shall adjust
Ka Lam LEE
20:44
Ok for me
Ying WAN
20:45
ok
Chu Wing Suet
21:01
ok
James Tung
24:52
yes
Ka Lung CHAN
24:53
YES
Ying WAN
24:53
yes
Yulin LI
24:53
yes
Ho Renee TIN
24:54
yes
Zhe CHEN
24:54
yes
Chu Wing Suet
24:55
yes
Zhe CHEN
29:01
yes
Ka Lung CHAN
29:05
YES
JIN Xg
29:05
yes
James Tung
29:07
yes
Ho Renee TIN
29:11
yes
Chin Ting Timothy TSANG
29:14
yes
Ka Lung CHAN
32:28
YES
Ka Lam LEE
32:29
yes
Chen CHEN
32:30
yes
Chu Wing Suet
32:30
yes
Zhe CHEN
32:30
yes
James Tung
33:27
Is it like Flattening?
Ka Lung CHAN
34:02
transform N X N to N^2 X 1?
James Tung
34:06
ok
Zhuo LIU
36:00
n*n
Ho Renee TIN
36:01
N^2*N^2?
Zhe CHEN
37:40
ok
James Tung
37:41
yes
Chu Wing Suet
37:44
yes
Chin Ting Timothy TSANG
37:44
yes
Zhe CHEN
42:13
ok
Ka Lung CHAN
42:13
YES
Chen CHEN
42:13
yes
Ho Renee TIN
42:15
Yes
Ying WAN
42:16
yes
Yijun CHEN
42:16
ok
Ho Renee TIN
43:58
Yes
Chen CHEN
44:02
yes
Zhe CHEN
44:03
yes
Ying WAN
44:06
Yes
Chu Wing Suet
45:55
yes
Zhe CHEN
45:56
yes
Daixi HONG
45:58
ok
James Tung
47:01
2*2
Zhe CHEN
47:10
2*2
Ka Lung CHAN
49:19
for M x N image, is H dimension M^2 x N^2 or MN x MN?
Ying WAN
49:54
M^2 x N^2
Ka Lung CHAN
50:39
YES
Chen CHEN
50:43
yes
Zhe CHEN
58:18
yes
Ka Lung CHAN
58:20
YES
Ho Renee TIN
58:21
G1=[1 0; 2 1]?
Yulin LI
58:21
yes
James Tung
58:26
ok
Ka Lung CHAN
01:01:10
YES
James Tung
01:01:15
yes
Ho Renee TIN
01:01:17
Yes
Zhe CHEN
01:01:17
yes
James Tung
01:05:18
so g is 3*3?
James Tung
01:06:00
ok
Zhe CHEN
01:07:36
yes
Ka Lung CHAN
01:07:37
what will be the dimension of f * g? N+2 x N+2?
Ho Renee TIN
01:07:38
Yes
Ka Lung CHAN
01:08:45
Yes
Chin Ting Timothy TSANG
01:11:41
ok
Ka Lung CHAN
01:11:44
the boundary will be do with zero?
Ka Lung CHAN
01:12:18
yes, zero padding at the boundary?
Ka Lung CHAN
01:13:03
so the -1 row will be the N row
Ho Renee TIN
01:13:08
It is like when we try to recover the noised image, we want to use the intensity of the neighbourhood pixels to estimate to centre one?
James Tung
01:13:11
Do we overlay a 3*3 or N*N window in this case?
Ka Lung CHAN
01:13:58
yes
James Tung
01:13:59
ok
Ka Lam LEE
01:15:52
Do we need to divide the sum by 9 when we are talking about average?
Ho Renee TIN
01:16:06
Thank you
Ka Lam LEE
01:17:14
Ok, got it. thanks
Ka Lung CHAN
01:18:14
YES
Zhe CHEN
01:18:18
yes
Chu Wing Suet
01:18:18
yes
Ka Lung CHAN
01:20:24
YES
Chu Wing Suet
01:20:24
yes
Chen CHEN
01:20:29
yes
Chin Ting Timothy TSANG
01:21:25
wow
Zhe CHEN
01:22:40
Why this happen?
Ka Lung CHAN
01:22:42
YES
Chin Ting Timothy TSANG
01:22:42
why -1 and 5?can it be -2 top bottom left and rightand 9 in the centre?
Tong WU
01:22:43
why the brightness lower?
Ho Renee TIN
01:22:51
But It's darker too
Chin Ting Timothy TSANG
01:24:38
i see
Ka Lam LEE
01:24:54
What if we do the convolution say 10 times?
Chin Ting Timothy TSANG
01:27:22
blur?
Ho Renee TIN
01:27:30
a little bit blur
Ho Renee TIN
01:30:20
But we shall be able to detect the edge. can we first separate the edge and the remaining image, one do shapening and the other one do denosing?
Ka Lam LEE
01:32:09
ok
Chin Ting Timothy TSANG
01:32:09
yes
Zhe CHEN
01:32:10
yes
Ka Lung CHAN
01:32:11
YES
Chu Wing Suet
01:32:13
yes
Ho Renee TIN
01:32:15
yes
Ka Lung CHAN
01:38:52
yes
Zhe CHEN
01:38:53
yes
Chen CHEN
01:38:53
yes
Ka Lung CHAN
01:41:06
YES
Chu Wing Suet
01:41:07
yes
Chen CHEN
01:41:08
yes
Zhe CHEN
01:49:20
yes
Ka Lung CHAN
01:49:22
YES
Chu Wing Suet
01:49:23
yes
James Tung
01:49:26
yes
Yulin LI
01:49:27
yes
Ho Renee TIN
01:49:29
Yes
Ka Lung CHAN
02:11:11
YES
Chu Wing Suet
02:11:13
yes
James Tung
02:12:28
matrix multiplication
James Tung
02:15:24
yes
Zhe CHEN
02:15:25
yes
Ka Lung CHAN
02:15:25
YES
Ka Lung CHAN
02:23:37
N x N?
Zhe CHEN
02:23:41
n*n
Ka Lung CHAN
02:27:21
YES
Chen CHEN
02:27:23
yes
James Tung
02:27:24
yes
Chin Ting Timothy TSANG
02:27:28
ok
Ho Renee TIN
02:29:45
yes
Zhe CHEN
02:30:31
n^4?
Ka Lung CHAN
02:32:18
CAN WE FORCE U AND V BE INTEGER?
Ka Lung CHAN
02:34:07
in bitmap format, the pixel are in integer?
Ka Lung CHAN
02:34:57
because integer will use less space and no loss in doubole
Ka Lung CHAN
02:35:34
OK. Thank you.
Ka Lung CHAN
02:36:02
will there be any condition that the image cannot be decompose? e.g. something like non-invertible
Ho Renee TIN
02:36:02
Yes
James Tung
02:36:03
yes
JIAJI LIN
02:37:04
SVD
Ka Lung CHAN
02:39:57
OK. Thank you.
James Tung
02:40:05
yes
Zhe CHEN
02:40:05
yes
Ka Lam LEE
02:40:05
yes
Ka Lung CHAN
02:40:55
I think SVD can decompose all matrix
Ka Lung CHAN
02:41:57
OK. Thank you.
Ho Renee TIN
02:53:51
So the other norm have no specific usage in image processing?
Ka Lung CHAN
02:53:52
Yes
Ho Renee TIN
02:54:15
oh
Ho Renee TIN
02:55:44
Thank you
Ho Renee TIN
02:55:49
that's interesting
Ho Renee TIN
02:56:43
we didin't prove the commutative property of convolution?
Ho Renee TIN
02:56:56
oh
Zhe CHEN
02:57:01
Can I see the transformation matrix as M*N transformation matrix in linear algebra?
Zhe CHEN
02:58:26
thanks
Ho Renee TIN
02:58:48
Thank you
James Tung
02:58:52
Thanks
Ka Lam LEE
02:58:54
Thank you and good night
Ka Lung CHAN
02:58:54
Thank you.
JIN Xg
02:58:56
thanks
Ronald LUI
02:58:57
Thank you very much for attending this lecture.
Chen CHEN
02:58:57
thank you
Chin Ting Timothy TSANG
02:58:57
thank you