0:00:13 | so um |
---|
0:00:14 | my name is a somewhat what an bit from japan and |
---|
0:00:17 | so today i would direct to talk about uh about my |
---|
0:00:22 | a a a little work |
---|
0:00:24 | uh tight so that all |
---|
0:00:26 | fast identification of jpeg two thousand images for distance there's to provide |
---|
0:00:34 | uh |
---|
0:00:35 | this is a i'll try all of my talk |
---|
0:00:39 | part a i'm direct talk about C |
---|
0:00:42 | but grand all of this work |
---|
0:00:44 | a second |
---|
0:00:45 | yeah |
---|
0:00:48 | jpeg two thousand coding system |
---|
0:00:50 | and it's coding the power as uh some rice |
---|
0:00:53 | then |
---|
0:00:54 | then |
---|
0:00:55 | the calm an image identification miss out of what chip pick two so than |
---|
0:01:00 | images |
---|
0:01:01 | is introduced |
---|
0:01:03 | and in |
---|
0:01:04 | the program all was calm based a miss not uh describe |
---|
0:01:11 | the procedure zero |
---|
0:01:13 | a missile to comes next |
---|
0:01:15 | for the by |
---|
0:01:16 | experiment in |
---|
0:01:17 | results |
---|
0:01:20 | and if |
---|
0:01:22 | yeah i it's at time |
---|
0:01:23 | i we conclude my |
---|
0:01:25 | what |
---|
0:01:31 | um |
---|
0:01:32 | did does him |
---|
0:01:33 | uh is |
---|
0:01:35 | no no as |
---|
0:01:36 | one oh |
---|
0:01:37 | uh |
---|
0:01:39 | the most successful up of case one all say pick two thousand |
---|
0:01:44 | uh jpeg peg two thousand has a coding |
---|
0:01:47 | provides |
---|
0:01:48 | for just as cinema application as an i mean doesn't and all that to standard |
---|
0:01:55 | uh in this |
---|
0:01:56 | provide a uh coding parameters |
---|
0:02:00 | uh all typical two thousand made if a |
---|
0:02:03 | for |
---|
0:02:04 | it's |
---|
0:02:04 | provide |
---|
0:02:06 | um |
---|
0:02:07 | in this just name and applications |
---|
0:02:10 | uh |
---|
0:02:11 | such as a |
---|
0:02:12 | and being or |
---|
0:02:15 | we in |
---|
0:02:16 | and so on |
---|
0:02:17 | uh there are needs to |
---|
0:02:20 | i didn't to find and exact frame |
---|
0:02:23 | exact the same frame which has the same or is no image as a given or is not image |
---|
0:02:29 | uh |
---|
0:02:31 | this identification called image identification and uh |
---|
0:02:35 | this identification of jpeg two sounds and images are |
---|
0:02:39 | required a it's its S name because uh |
---|
0:02:42 | frames |
---|
0:02:43 | all |
---|
0:02:44 | this that's the contents |
---|
0:02:46 | uh encoded by a typical two sets and |
---|
0:02:50 | so um |
---|
0:02:52 | image i didn't |
---|
0:02:53 | you meet identification is |
---|
0:02:55 | defined |
---|
0:02:56 | uh an operation |
---|
0:02:58 | for finding an image identical |
---|
0:03:00 | to a a given the image from you may database |
---|
0:03:04 | and that this |
---|
0:03:05 | processing should be done in compressed domain |
---|
0:03:10 | uh |
---|
0:03:11 | how about uh commissioner miss thoughts |
---|
0:03:13 | cannot handle the chip to send images support just them |
---|
0:03:18 | um because |
---|
0:03:20 | in this this smell |
---|
0:03:21 | provides |
---|
0:03:22 | uh coding parameters such as a |
---|
0:03:24 | the average a label on |
---|
0:03:28 | and that quantization step size and the call broke side or or and so on |
---|
0:03:33 | um may differ |
---|
0:03:37 | um |
---|
0:03:38 | now i would like to talk about jpeg two thousand coding system |
---|
0:03:45 | this is a a diagram or jpeg two thousand coding system |
---|
0:03:49 | and uh |
---|
0:03:50 | decoding power meters |
---|
0:03:52 | using and jpeg two says uh used in jpeg two thousand |
---|
0:03:56 | uh according to each operation broke |
---|
0:04:00 | uh a was that rate character |
---|
0:04:03 | so uh |
---|
0:04:06 | i |
---|
0:04:06 | fast uh input image is uh |
---|
0:04:09 | and i Z by D that would be and think ties it and they're |
---|
0:04:12 | them |
---|
0:04:14 | uh each |
---|
0:04:15 | transform questions uh |
---|
0:04:18 | divided |
---|
0:04:19 | uh called rock |
---|
0:04:20 | and then |
---|
0:04:22 | bit about it meet the ring |
---|
0:04:24 | a |
---|
0:04:27 | uh it's down |
---|
0:04:28 | so after a |
---|
0:04:29 | bit modeling |
---|
0:04:32 | the number of bit of brains |
---|
0:04:34 | uh |
---|
0:04:36 | it called in this presentation |
---|
0:04:38 | a a and and that B P |
---|
0:04:40 | uh |
---|
0:04:41 | is produced and the recorded it as a part of the had that information of jpeg two sounds and called |
---|
0:04:47 | the story |
---|
0:04:50 | uh in this work |
---|
0:04:51 | uh |
---|
0:04:52 | this |
---|
0:04:53 | and that the bp |
---|
0:04:54 | is used to fall image identification |
---|
0:05:00 | uh |
---|
0:05:01 | i would like to talk uh |
---|
0:05:03 | that that detail four |
---|
0:05:05 | uh and that P P |
---|
0:05:08 | uh this figure shows uh |
---|
0:05:10 | transformed questions |
---|
0:05:14 | uh |
---|
0:05:16 | this is a wherever it questions and the |
---|
0:05:19 | to this it's a seven one |
---|
0:05:21 | and uh |
---|
0:05:23 | the rocks |
---|
0:05:23 | uh here |
---|
0:05:25 | so called rocks |
---|
0:05:27 | uh |
---|
0:05:30 | each quantise questions is |
---|
0:05:32 | separated into this |
---|
0:05:34 | each sign |
---|
0:05:36 | and the i was sort it should |
---|
0:05:39 | and the |
---|
0:05:40 | where a |
---|
0:05:42 | we have the absolute to use of |
---|
0:05:44 | for tries |
---|
0:05:47 | a speech prince |
---|
0:05:49 | from the most significant bit to say the least significant country |
---|
0:05:54 | all the samples in the bit prince uh |
---|
0:05:57 | is out |
---|
0:05:59 | zero or one |
---|
0:06:03 | then is a special breed brain |
---|
0:06:05 | uh in fish all samples are zero |
---|
0:06:10 | it's a P S |
---|
0:06:12 | and this special bit-plane is if to as a |
---|
0:06:16 | zero a brain |
---|
0:06:18 | and then when all |
---|
0:06:20 | it prints out |
---|
0:06:22 | zero be brains |
---|
0:06:23 | uh the cold broke |
---|
0:06:26 | this |
---|
0:06:28 | uh this is i |
---|
0:06:29 | coding you need to a chip pick two cells and |
---|
0:06:31 | uh |
---|
0:06:35 | defined as a not included |
---|
0:06:37 | broke |
---|
0:06:38 | because it does not include any data that to be encoded |
---|
0:06:52 | uh this figure shows the |
---|
0:06:54 | example all the |
---|
0:06:58 | and P P |
---|
0:07:04 | and not that uh and that B B |
---|
0:07:07 | can be easily extracted by passing to hit that all jpeg two thousand called streams |
---|
0:07:13 | and uh |
---|
0:07:13 | no know coding is it |
---|
0:07:16 | and uh compression late |
---|
0:07:18 | or was it called a stream |
---|
0:07:20 | has no impact to |
---|
0:07:22 | to bad in that P P |
---|
0:07:25 | uh there is a |
---|
0:07:27 | metal |
---|
0:07:27 | two |
---|
0:07:29 | vol you mean image identification |
---|
0:07:32 | using |
---|
0:07:32 | and that the P P |
---|
0:07:34 | uh i would like to summarise uh miss a |
---|
0:07:37 | uh this ought passed to extract the |
---|
0:07:40 | and that to be P using a |
---|
0:07:42 | hit the se |
---|
0:07:44 | and then |
---|
0:07:45 | uh |
---|
0:07:47 | the components and all in to be P between a |
---|
0:07:51 | query eight |
---|
0:07:52 | big two thousand images and the i image if |
---|
0:07:55 | mean |
---|
0:07:56 | and the J peg two that the images in uh |
---|
0:07:59 | and you made database |
---|
0:08:07 | uh this mess of works where for the a you is that to have the same coding power as |
---|
0:08:14 | however about it is not applicable with |
---|
0:08:17 | different coding parameters |
---|
0:08:19 | because uh |
---|
0:08:22 | barry all and that the P P in images |
---|
0:08:25 | compressed to is different coding parameters may be different |
---|
0:08:33 | so two so was uh |
---|
0:08:37 | this program |
---|
0:08:38 | uh the extended |
---|
0:08:40 | ms of it's probable |
---|
0:08:46 | uh the proposed missile |
---|
0:08:48 | a is composed of four steps |
---|
0:08:50 | at fast |
---|
0:08:51 | uh regions to be compared to a defined |
---|
0:08:56 | using the |
---|
0:08:57 | regions |
---|
0:08:58 | uh the difference of the labels and the or |
---|
0:09:03 | code |
---|
0:09:03 | called broke size can be can say |
---|
0:09:08 | uh here is a |
---|
0:09:10 | transform the sent |
---|
0:09:12 | uh of a query image |
---|
0:09:14 | and a this is the database one |
---|
0:09:17 | and uh |
---|
0:09:18 | each |
---|
0:09:19 | a a query images |
---|
0:09:21 | uh be composed to by W dwt |
---|
0:09:24 | and level four |
---|
0:09:25 | and don't data bases a three |
---|
0:09:28 | and we choose uh |
---|
0:09:31 | many |
---|
0:09:32 | little |
---|
0:09:33 | between the |
---|
0:09:34 | images |
---|
0:09:36 | uh |
---|
0:09:37 | for for identification |
---|
0:09:39 | so |
---|
0:09:41 | this is just have a one |
---|
0:09:43 | so |
---|
0:09:47 | in could we made this uh the rocks size is set to |
---|
0:09:51 | a to by set to |
---|
0:09:54 | but uh in |
---|
0:09:56 | database images uh |
---|
0:09:58 | that the |
---|
0:09:59 | uh for group size |
---|
0:10:01 | uh it's set to sixty four by sixty four uh |
---|
0:10:05 | they are |
---|
0:10:06 | define in each other the |
---|
0:10:09 | uh |
---|
0:10:13 | the whole so is on top and the but can sides off |
---|
0:10:17 | compared regions |
---|
0:10:19 | uh |
---|
0:10:19 | sit to user |
---|
0:10:21 | recent |
---|
0:10:22 | marks marks spot |
---|
0:10:24 | or |
---|
0:10:28 | this site |
---|
0:10:30 | so uh |
---|
0:10:32 | this is a example compared regions |
---|
0:10:37 | uh |
---|
0:10:38 | one regions |
---|
0:10:39 | uh |
---|
0:10:40 | has a a a a full whole blocks |
---|
0:10:43 | for a query images but uh |
---|
0:10:45 | one regions has a a one of the broke for database minute |
---|
0:10:52 | uh if |
---|
0:10:53 | but blue |
---|
0:10:55 | number of did zero bit to in the region like a query images |
---|
0:11:01 | uh or is the minimum value is used |
---|
0:11:05 | in some so you can't |
---|
0:11:06 | process |
---|
0:11:09 | like is |
---|
0:11:10 | uh a few days |
---|
0:11:12 | not included |
---|
0:11:13 | but you |
---|
0:11:14 | uh minimum value |
---|
0:11:17 | is you priest the by |
---|
0:11:18 | not good |
---|
0:11:24 | and then uh |
---|
0:11:26 | omission mission or |
---|
0:11:27 | relation vectors |
---|
0:11:28 | uh |
---|
0:11:29 | a right |
---|
0:11:30 | so the |
---|
0:11:31 | and that B P all joe for a region |
---|
0:11:36 | this is our region |
---|
0:11:39 | a region |
---|
0:11:40 | and that it's |
---|
0:11:41 | okay |
---|
0:11:42 | neighbouring regions are can be a and then |
---|
0:11:45 | so a back that is |
---|
0:11:47 | a produce |
---|
0:11:52 | this value all |
---|
0:11:54 | one |
---|
0:11:55 | means a in that B P |
---|
0:11:57 | or was a neighbouring regions is |
---|
0:12:00 | small runs and |
---|
0:12:02 | for or one |
---|
0:12:03 | and the minus one |
---|
0:12:05 | means a a and that to be also a neighbouring regions is run and |
---|
0:12:10 | folk or one |
---|
0:12:12 | and uh |
---|
0:12:14 | the components comparison |
---|
0:12:15 | with |
---|
0:12:16 | anybody value and or |
---|
0:12:18 | between uh |
---|
0:12:19 | not to include it |
---|
0:12:21 | oh to put |
---|
0:12:22 | zero eight minute |
---|
0:12:24 | all was a big to |
---|
0:12:27 | oh finally the local decision |
---|
0:12:30 | uh |
---|
0:12:31 | is made by these operation |
---|
0:12:37 | this local decision it's four |
---|
0:12:39 | one |
---|
0:12:40 | compared regions |
---|
0:12:42 | and the when uh |
---|
0:12:43 | for or |
---|
0:12:46 | uh |
---|
0:12:47 | this is a is |
---|
0:12:48 | or all or all or can the regions |
---|
0:12:51 | uh |
---|
0:12:53 | but steve |
---|
0:12:55 | this shows are made |
---|
0:12:56 | so |
---|
0:12:57 | the global but this is a is |
---|
0:12:59 | may |
---|
0:13:01 | if all |
---|
0:13:02 | local decisions uh point steve and then |
---|
0:13:07 | colour |
---|
0:13:08 | global but this isn't is |
---|
0:13:10 | plus still |
---|
0:13:15 | in the experiment uh we use the |
---|
0:13:19 | standard the evaluation just marked area uh it is defined as a uh |
---|
0:13:25 | D C i'm |
---|
0:13:26 | uh this never and you said tape |
---|
0:13:28 | uh was used |
---|
0:13:30 | uh |
---|
0:13:30 | this just it |
---|
0:13:32 | contained a uh over |
---|
0:13:36 | fifteen thousand frames |
---|
0:13:38 | and uh have fast |
---|
0:13:40 | a sequence of jpeg two thousand compressed |
---|
0:13:44 | you may is was to in a database |
---|
0:13:48 | and uh a query images once found a and uh and in five identified to an the jpeg two thousand |
---|
0:13:54 | images in that data |
---|
0:14:00 | uh the identification experiment how was performed to for all possible combinations of a query and a database images |
---|
0:14:10 | and the number of combinations of a query |
---|
0:14:13 | and the database |
---|
0:14:15 | well as a |
---|
0:14:16 | fifteen |
---|
0:14:17 | so was and by fifty and in total |
---|
0:14:25 | yeah it's a property of of the |
---|
0:14:27 | test C against and then the encoding power we can support query images |
---|
0:14:33 | uh the resolution |
---|
0:14:35 | all |
---|
0:14:36 | this C case |
---|
0:14:37 | is a |
---|
0:14:38 | four sounds and by |
---|
0:14:39 | two thousand |
---|
0:14:41 | and uh |
---|
0:14:43 | this is a |
---|
0:14:45 | so as as a coding a permit to as |
---|
0:14:48 | so and the |
---|
0:14:50 | this table |
---|
0:14:51 | a a is the power as support database images |
---|
0:14:55 | in this table |
---|
0:14:56 | uh |
---|
0:14:57 | this |
---|
0:14:59 | did to you piece if you've |
---|
0:15:01 | five three and C B |
---|
0:15:05 | C six people C B want to eight |
---|
0:15:07 | are we is |
---|
0:15:08 | i tape |
---|
0:15:10 | uh means uh images |
---|
0:15:12 | encoded using a different |
---|
0:15:15 | do you double D V you as and the core the block is |
---|
0:15:18 | a divide to iteration and different step sizes |
---|
0:15:26 | uh |
---|
0:15:27 | this is an experimental results |
---|
0:15:29 | and the |
---|
0:15:30 | the |
---|
0:15:31 | is force |
---|
0:15:32 | was you |
---|
0:15:33 | right |
---|
0:15:34 | if B R and the two to rate T B R |
---|
0:15:36 | is |
---|
0:15:37 | uh kinda kinda correct |
---|
0:15:41 | by |
---|
0:15:42 | these equations |
---|
0:15:47 | this table shows so |
---|
0:15:50 | the idea of identification based on the proposed a of produce to and uh |
---|
0:15:55 | one percent |
---|
0:15:56 | all |
---|
0:15:57 | the |
---|
0:15:59 | a a one percent |
---|
0:16:00 | if B are we of the difference of jpeg two thousand coding probably |
---|
0:16:06 | and the result any false negative match |
---|
0:16:15 | and uh |
---|
0:16:16 | this experiment |
---|
0:16:18 | a X Q to |
---|
0:16:20 | uh |
---|
0:16:22 | on a workstation with |
---|
0:16:24 | uh wrecked is |
---|
0:16:26 | and that the is processing time for one combination of a curry and uh |
---|
0:16:31 | database a image was about uh |
---|
0:16:34 | the point six |
---|
0:16:35 | me second pass frame |
---|
0:16:40 | a a a a finite i really |
---|
0:16:42 | i would to conclude the my talk |
---|
0:16:45 | uh |
---|
0:16:47 | and identification method of what jpeg two thousand images with uh |
---|
0:16:51 | different jpeg two thousand coding parameters has be |
---|
0:16:54 | present it |
---|
0:16:55 | uh |
---|
0:16:56 | for each as |
---|
0:16:58 | uh |
---|
0:17:00 | the proposed wasn't miss out |
---|
0:17:03 | is uh |
---|
0:17:05 | compressive domain processing |
---|
0:17:07 | and uh has a robust S against a difference |
---|
0:17:11 | encoding calling S |
---|
0:17:12 | and uh |
---|
0:17:13 | it's one |
---|
0:17:15 | runs |
---|
0:17:16 | last |
---|
0:17:17 | thank you for your attention |
---|
0:17:24 | any questions |
---|
0:17:31 | i i i one push |
---|
0:17:33 | i you should in the previous slide that but the uh i |
---|
0:17:38 | it T P R of the true positive |
---|
0:17:40 | and it was at a hundred percent |
---|
0:17:42 | does it mean that you could actually get results that the overall better but proving the F P R |
---|
0:17:48 | introducing reducing P are a little bit |
---|
0:17:51 | if you change the threshold |
---|
0:17:53 | a a actually a a |
---|
0:17:56 | we |
---|
0:17:57 | we have a |
---|
0:17:58 | we have a |
---|
0:18:00 | for home to some experiments uh a with a a different |
---|
0:18:05 | coding rate |
---|
0:18:07 | so we get the |
---|
0:18:09 | very low rate |
---|
0:18:10 | uh some |
---|
0:18:12 | results a |
---|
0:18:14 | uh oh obtained |
---|
0:18:16 | uh |
---|
0:18:18 | not hundred in |
---|
0:18:19 | "'kay" |
---|
0:18:21 | but uh |
---|
0:18:23 | oops |
---|
0:18:23 | also so |
---|
0:18:25 | much of different |
---|
0:18:30 | a a and you the questions |
---|
0:18:32 | Q let's the to speaker |
---|