that's one

um i um is and the

a pair of uh

uh

but subject to

change

are

but

the some

uh the more

it's

yeah

for

true

yeah

also

uh the uh

modeling issues

i

start i want to go through a very quickly through the enjoyment you have an image

by which the what we call excel

uh they stay

on a

i

i

is it two

a is

and do can excel as that was sent there's

a

that time was you see you know

uh

that

or noted the U are

two

a so are

such as

so

and our goal is to

uh are present the contour of

a each

you go by image it has but use

for while

zeros

model

and that the ones are

yeah

for that foreground this zero a the debate

the court would stay in between the for foreground and background

if by to be contained in the foreground the cells

not to you want to be three

for you see there

for mean required for

binary image

they have the property that uh

uh

they have an a part which is on the big

so a the so from the

for a each has an they work on them

the ground

would be a don't to the core

uh

the the a being counted he you know

for one it

so what we can see that the

a binary in each and the contour for are

or

we in the same information they can be a

no one one mapping whatever you have one we can

you i the other as well

so called or more in a binary image is you to call or modeling a whole

yeah going to a first of one for the sequence of big sense you one N

each chi a a a it's so for an A the would be given

mentioned

like T

and

a property P of this change of excess would be that to

uh and and pair can see that you the it sense would be connected in eight

i

or are for the creation

uh which you have been

yeah

that is that

and the court it's of the next uh

uh

so in the for it's of the last but i G placement

which are a a lot of be in as one

one

set

except they uh

but zero zero so we

don't allow here

so

to state

a things

we have always move

a but i right all the directions

right and north

so it that that are not R

no you if we look at all these past what three uh

movements months on one

for

we can see that a uh we can buy eyes

this

very is you by a number you

which has a they use from zero to seven

and the equation

the

to get

the court it i G from

uh you think about a phrase

a number of you

would be

but i mean

the course

i i times

that that five

or four times

and the same uh that for sinus

you of the or the G

and you we have that dog

so now we make all these uh

a was a kid eva

uh

mark was of five or for

they are going to get the allowed

oh zero would be

for for

well be back don't

uh

down works

uh then two

but what or and so on so all of these are allowed

but and this a ah a prioritisation uh

it does uh introduced in that

this segment is the

a few it can be cheap

which uh uh

so that the

we have a part their

that's zero seven

and

uh our final change own would be a sequence of

those that but

don't numbers

can be a uniquely defined by

is you quality so if we take a job you got uh

one issues of the pixels

but

that's lot

i T

we get the

i i to me

at time

a a a a that is a very simple because uh fashion

to get the job but you presentation based on that

uh on there

a now we can component that

we have

a but of the point by

you

the first position of the

so

say i'm core

chain

and then a sequence of a

simple

the change

sequence is now be in this set zero you know seven

and that would be

no one one correspondence to the gym you can

sequence of excess P

you want to P M

no i was observed that the

uh uh this uh uh a few might quotes are would be to D dot then and it's much more

a efficient two

introduce a new variable called the bit at T

might as well

when

the and cold

but actually D uh a my

simple so we have

but minds

the might as well

that would be

we that is like we take a is operation

uh we

but as the model at uh

subsection

and they they the non-negative in

is a uh

now we see that uh

these new numbers be tough

uh are also connected to the positions

because of each other

but your are uh called binary T

and in fact is that there is a very simple and nice the passion for getting the positions of pieces

the second point

you the position of the peaks the previous time

and the

uh

next to used in the box

and are

new i'm then be thirty

which is the partition

you you might change

where these metrics is that methods similar

so now i mean that a a second uh it really representation of V C of pixels

that the position of the first six and what addition of the second um

and then we have that

uh a chain code

you can be numbers in this

said the

uh

okay and it's in being model they we have a number this at zero one up seven

yeah is um

uh

to code we you discuss a now it would

you a are you to be seven that was that B

point

where you have to move

as you see they are a

a when T V D so we move sometimes that all

like here

sometimes

because sometimes

okay that but

it is what

well as uh

it's so wise

it

i Y way to represent the binary make

but there is also a lot and i think we how to

we presented

the the by

introducing

yeah

and S

got that P which are

packages are is are to those

so they are got three

use

the back is these be uh that this is of those at

so we can uh uh also discuss about the chain codes are representations or water

just like it is that is

to now we have been in the this that this points you want to and so on

if we translate

the that is by a step

vertically and horizontally we are going to get a like things and

we can discuss about change "'cause" when this describe that

but is interesting is that now we have to move just in for what a T V so this uh

but this is

i mean to provide a something in for it but

and the um

we see that the uh

length

of the joint code here would be

use a number member and we went much longer than the change a the peaks

do have just seven except the J

you would have sixteen

now the question is

which would be the best presentation which one

should be used for encoding

uh the same P and and T which is the one of the age

that it is and exactly the same information so

we have to use

uh which are better

use the best is result

a discussing about was so uh but that uh

a chain codes

we can do to this exactly as before

uh

that

female

chain code in the uh what if D V D for

where you have a

a very similar to the definition as a for now this numbers almost

a a in this a zero one two three

that just for values for grammar

and uh our

what the image you would be no if we want and

uh two D set the initial that X

plus the

yeah my one uh points from the

uh a change words on the task with that

we can two also of attention

uh codes by introducing these difference between galactic that one runs on

that would believe that it is gonna before

that is

for this is that uh uh it is that is that is not just think that

i whole

which is quite a few O T

are introduced there

later than

well yeah that's

is is that to the fast at the

uh you uh of you well

but not it's not

yeah

for for that

a definition my

the difference between two uh

you my quotes

you

these

second order

then memory

now we don't have any more a second order memory more in fact we have a model with the

uh you put a very long memory

uh because now introduce to state variables

which i we to tell if

the sense of

a movement or along the horizontal vertical direction

uh

i get or not so he's are state that

that would be to state tiger was

and the code itself would be defined as

uh is you know in that is no change direction

why if there is a change that but but you that

well sense

whatever whatever moved that direction

and the call to would be one you change point detection and the sense

no that direction

so this would be a very uh

lot of memory of the nation and i for now we had a have to say but

and what of be

chain code then what is the next

that that's in the chain of and

sequence of

uh

yeah a representation for is very simple going to with that we have the corporate along that the they sounds

there

thirty seven cents

and the colours are in the set that zero one to

up to seven

and and you have also the called uh

a i the got yeah images

and

you see this sequence of uh

simple in the

a a T a but as you one and two

uh we have a much more C press of that this is a you T all of them

combat to thirty seven

except

and that may say that probably this is less efficient cool

but is not

and right as in is that uh

we have a few force what but they are just zero one and two

but a more uh uh what is more interesting is that the simple to

we you know

change

direction and sense

but you is that we sell them in the scene

so basically they we we have

to see then it to see what it should get most of the time

and uh

this would be a

but it should be able

what would be to for uh why we are to see this this the compression problem

uh uh because the

uh

you may want think one by reading is and then would be important

so

but may also can here

a very good statistical models

object

we can

call this

features

objects

for a a recognition

and main wants to know what is the minimum description length for

uh

this models

so for that reason

we are interested in the most that's compression of the contours and that's who the study with need in this

the they

well so there is not but um which is but yeah well what did but is not identical is that

notation

contours in image segmentation and image segmentation also

a very important the uh a region and you do we need other

you

what is the state of that the of corporation of four chain codes

uh how how a codes have been used to the twenty years ago um

they are a fixed

uh

codes

defining a simple was for the for a the extensions for who but that was uh used the five years

ago

a a able to variable able uh coding run it's building that

also use

uh a what was used in a very nice paper in two thousand six but the recognition was uh a

or text these for

this code

the

you might

and that was used for map uh

what of mess

they have a C are they are they specific or

uh a lot of this i i think of it to um

conflicting here

a results

not i is this is it telling you should use this uh uh them a the back pages follow

like to publish and you what

the best these definition

if a code

the conclusion in this paper would be that

this is not the best is second best but the the best but would be in that a a three

T which

is more

that's one reason why it was not start it so well not

so the question would be what is the best

the should should more than four

or the modeling

we that was something in important this type

conditional probabilities what would be that the probability of

the key even all that D

the previous moment

and we have the same for big so and here for but

so our goal is to a which of these is that

better suited for our

one want to

a

a

uh i our um

first for the second are model we we have just introduced at these few my codes and iteration not well

but in our paper but we are we can get

also a a higher order models

what you combine these first order and second order models

uh in

uh

my model so

memory

a which use a fixed but next memory

or or you models where the memory of the model depends on a

so we have a or or just uh compressing a the point two

for is the

that's pretty

oh time series is then

we product working

to you should use in the decoding and have a

this is a um

need to um is that three where we have a

on the route

or one on the

nodes

the nodes

we have

deviations

oh

we have to an estimate notes that was a channel

uh they can either or distribution

that was which are thirteen

they are used

ooh

when whenever code so we joined the distribution quality

these note

now

when it would coding

in a lot of this court

so this kind of a a context you recorded data

we context the

i i assume is not uh

what did you are but they do find and you that of the digital

uh a a a a base the T but uh

not is pulling

it was

less uh

uh uh use how they it the the the coding is that paper mission

by a a over and the from

used for depression

my

a a big problem of be how to optimize the point T structure and we have but

class of that i a problem uh

for choosing this uh

structure of the T that's

any

one context you

which is four

uh

symbols in the previous positions in the G

um

we put that the statistics of of of of the counts we do i'd made in here so

well the time we collect at the T V what would the code lengths at time T

but the index

but that's in that's

and uh

we are it mostly in the find a a a a a get to the and the

of

the the sequence

get that

and uh uh are study after we get to the at least i be at each node what we

what is the balance between coding

don't

coding to the choose

we just at the codes of the children or back to the bottom

you for be a a part and is uh

next year is that than the children last

a T

uh

because you have to encode the structure of this uh

a those relations

and whenever a this happens uh we you this you know as and that you node otherwise

it would be a as a uh

i

and we provide like data

five or or for this next uh

politicians

these calls

that is that i think that um

that and program

a way to

optimized the quantity structure

now we've the putting we have a a bit more

did you got or more more thinks to

study we um

like the statistics not only that the

note to and time T

well so we take all the possible groupings of E

children

a second no

and and a complete the

such a a a code

oh that is

because the same so

choosing which two

to use would be the uh

a a class the lucky

which is now

we can do that

uh

structure of the three

so that is a a set to me a a a a are adapted the

a we encode

first thing for the structure of the T we number that is the best the at and of our change

all

and then we just we that uh chain code the with a i keep statistics but this that no no

a

or

two show some results

but that we have this a set of the what high by use only when B are shown here

that a very different from you you know like to make it to you know but

a a some images is making many holes

uh i think you're

going

we can

everything think there which uh

would be equivalent to a won't the full binary

one example is this uh a set in each

and you can see what is that the optimal structure of the P

for a

uh you T change go

and that would be to what the most structure forty uh to finish if you michael

this uh

uh uh if it should if you might what would have just a five song

the

for i

simple this one seven

something to hide that

the structure of the fee

is you more interesting if you represent dramatically what would be a to "'cause" that these of this call

you we have the context

what do with the excel

you have the context text up with a a a guy it is

and you that the german because sectors are no longer for the crack them for the because

although the two are

it but and representation

the context and this statistical on uh a it's to and S is different in the two representation

maybe i'm going the desired

yeah the results for the one having five we had in the that sense

yeah a and we uh it represents in um

course

or the performance of the

and probably by a context C everything here

we different from the a chain codes

the that really to be as your T which is the bands

well yeah or so that would be the zero line

oh yeah as is that you can see a higher on top of these you don't my meaning that they

are not as efficient as the code you want the and that happens for all the five

the best company but is still on the uh

yeah a is the core the

a a month with this

neighbour

for

uh and

then um

blue um is there uh

the best the context the on T D

excel one

is

a that that was a is i it's of a a code by grouping

i this is that the the sure that is that that's which are published in a recent paper in uh

at

yeah

just of that since the processing

but that is as well as you can see here

i results in the context you at what is the

i a much better

a

tens of course and

a a better than

then the

this one

uh here what they use that you just at a few or more

more

and they established by a uh yeah you great again that the best legs for all the models would be

be

well as we have a uh

due to a possibility to take the longest uh one text

and the were used other markov model

and we you got but but a result

for

or T one hundred five

and for each individual files or

is that a special but is day

uh dependence of the possible and that's so be um

models

and as you see uh our T models

such a some points so at this point you may say that this they

what they're six

uh a of the models

so this the same that additional

it happens us by the same or

the same plane

as got images

we have many presentations ah a a of the contour data are uh we found that a statistical efficiency of

in E papers right but

change

and the uh when we have when we use uh a T context modeling for a

uh that is a significant difference between the trees for one in which what D image

so

for that we said that

but that we use in adaptive

but for

uh i think of

and what we see find it that the what we we won't and is not only

yeah uh important point coding but is also for a find good this kid of uh

binary images

i you the

i i i i have time to ask some session

for for a i

yeah