What It Would Really Take To Sink a modern Aircraft Carrier

[jhungary]

No, I am talking about using one software to differentiate between objects as varying as human and carriers! Then, even in that special features for each object, for humans, jawlines, eye position, nose, etc. For carriers, their silhouette, aircrafts on deck, wake, etc.
You said PRISM has been used for both the purposes. Quite difficult!
As for differentiating between Liaoning and Abe, what is the need to do that autonomously? In the entire world, there is less than 20 carriers, so if an algo calls human for max 20 times for verification, I would say it is an awesome algo!

I do not understand what you are talking about.....

PRISM needed facial recognition software to make out faces, and since PRISM cannot afford to be a manual system (Which would require many man hour to just monitor one subject) they will need an autonomic recognition function to do that.

It's not hard to separate human from aircraft carrier, as there are a lot of program even on the market can do this, but to perform a subset search, you need to write a whole new matrix for it, and depends on what you are comparing with.

 

[randomradio]

First of all, Miniaturisation is always important, you cannot compare the matrix if you cannot miniaturize or magnify the image, unless you limited the image input.

Second of all, the Liaoning is smaller than the Nimitz, yes, but how do the computer knows when they only can see shapes without miniaturisation?

The link I posted clearly said smooth approximations are used in lieu of minimization since threshold, or reference pictures are available.

Thirdly, about cars, as long as there are 1 UI, you can identify the car between smallest detail, given if you can capture that UI, the problem is can you capture them. I have worked with Traffic cam ID software before



lol, you do know how this program works, right? It does not recognize car, instead it recognize partial information (most likeky from the License Plate) from the image. You can see how the program works if you put in this photo in try me section

View attachment 393104

If you put it in the system, select automatic read number plate, then select rear view, then select A,D,C.........TK in the license plate option, it will give you.

Car (1 point)
VW (1 point)
Red (1 Point)

However, is the car red? In fact, the color is probably the most easiest part to get right by pixel matrix (each pixel have a different color code encoded in an image, something like FF00FF.

The problem now is how this system make this fundamental mistake? That can possibly mean only one thing, which the program is taught to read the license plate or something unique, then search the history (such as DSLV or some sort of motoring site) and find the detail that way.

It is further proven if you put an Australian car (With Australian Number plate) like this one

View attachment 393105

And it returned nothing.

You see how this work?

You missed the fine print. The link clearly says: The analysis is independent of number plate reading.
http://www.eyedea.cz/make-and-model-recognition/

This is about Automatic Target Recognition with invariant position, rotation and scale, as per the thesis pointed out

In order word, the scale, position and rotation must be constant, which mean it is a geometrical match.

The report only speaks of geometrical match, but it is unnecessary. You need data to expand on the report, which is easy enough when you are a military because only funds are the limitation.

You have confused the limitations of the report to limitations of technology. A satellite can take a million pictures from a million different angles. There is no limit.

Not a non-geometric match or dynamic match

The carrier dimensions are known. There is no ship in the world that has a beam that's more than 60m.

When you have a million rotations to compare with, your task gets easier. After all, there are only less than 12 supercarriers. Not to mention, you can watch the carriers constantly even during peacetime, which is determined by financial limits, something the Russians and Chinese do not have.

Although we have not considered this in the discussion at all, realistically think about it. There are only 12 carriers. You can assign 12 guys per shift to watch the carriers 24/7. As far as the Chinese are concerned, the ones in the Pacific are the most important. The only limitations are time on target and weather (for optical satellites).

Did you actually read the title and the article? It said NEAR REAL TIME. IN the article, it explained as 1 hours from target to end user

The limitation is processing.
Here's a quote from the report:
The detector runs fast (in minutes for 16873 × 14684 pixels image) on a modern multi-core computer, thus enabling near real time application, i.e. one hour from image acquisition to end user

This uses processing from a COTS PC, not the same as the processing the military uses through their custom designed workstations and supercomputers.

Plus, the targets used are very small, less than 20m length, so that adds to the delay.

See above. I wonder if you actually been to sea and/or to the air to know what you are talking about.

Most people who design/develop equipment have neither been to the sea or air. People learnt about air and sea after someone designed something that could go out there.

Newton didn't figure out gravity by falling off a building. He observed nature, just like most designers do.

It is not that difficult to visually confuse shapes and outlines of ships.

Weather is a problem for optical satellites, but not for radar satellites. Both will be used in conjunction.

You forget that the container ship will have the red, white and blue colours picked up by the detector.

This is a very low resolution of what the container ship will actually look like from space.

Even the dimension of the ship is available because each pixel is of a particular size, the accuracy is correct to centimeters.

 

[jhungary]

The link I posted clearly said smooth approximations are used in lieu of minimization since threshold, or reference pictures are available.

lol you do know that mean the result is shit, right? I mean, if they actually have a result.

You missed the fine print. The link clearly says: The analysis is independent of number plate reading.
http://www.eyedea.cz/make-and-model-recognition/

Then explain to me why when I put a Lancer EVO X in GB Plate

It generate a result, but when I put the same car in same color in Australian Plate

It then said it cannot be found?

Mind you, the GB result it generate is NOT AT ALL ACCURATE

Does this look like a Kia to you??

Just because it said it is independent from reading the license does not mean they don't, just that they say they aren't, otherwise, they would have pick up both car (the same make, model, year and color)

I will let you be the judge

The report only speaks of geometrical match, but it is unnecessary. You need data to expand on the report, which is easy enough when you are a military because only funds are the limitation.

You have confused the limitations of the report to limitations of technology. A satellite can take a million pictures from a million different angles. There is no limit.

You have yet to demonstrate that such technology exist now.

Even if a satellite can take millions picture of a subject in million different angle, it does not change the fact that it cannot process result in real time.

The carrier dimensions are known. There is no ship in the world that has a beam that's more than 60m.

When you have a million rotations to compare with, your task gets easier. After all, there are only less than 12 supercarriers. Not to mention, you can watch the carriers constantly even during peacetime, which is determined by financial limits, something the Russians and Chinese do not have.

Although we have not considered this in the discussion at all, realistically think about it. There are only 12 carriers. You can assign 12 guys per shift to watch the carriers 24/7. As far as the Chinese are concerned, the ones in the Pacific are the most important. The only limitations are time on target and weather (for optical satellites).

umm......what about MV Blue Marlin (Beam 63.2 meters)??

https://en.wikipedia.org/wiki/MV_Blue_Marlin

How about Dockwise Vanguard (Beat 79 meters)

https://en.wikipedia.org/wiki/Dockwise_Vanguard

A few cargo ship can have beam over 73 meters, here are the brochure

http://maritime-connector.com/wiki/ship-sizes/

A few ULCC and SLCC crude carrier is actually larger than an Aircraft Carrier with beam over 68 meters

https://en.wikipedia.org/wiki/TI-class_supertanker

Do you want me to continue?

Again, just because you said so, does not make it true...

and am I hearing this right? (Since there are 12 US Carrier in the world, then we only need 12 person to watch it 24/7??

How about Sat coverage? Are you for sure SAT coverage can cover all 12 carrier 24/7? From my first hand knowledge, US have about 4 times as much Satellite than the Chinese, we cannot have a constant surveillance for a target for over 2 hours in a day, you do know Satellite will go out of range, right? Because Satellite will orbit the earth. Either the earth move you out of the orbit, or your satellite went off the orbit itself. Either way, you can dream of a 24/7 satellite surveillance for a single target, let alone 12 in different corner in the world.

The limitation is processing.
Here's a quote from the report:
The detector runs fast (in minutes for 16873 × 14684 pixels image) on a modern multi-core computer, thus enabling near real time application, i.e. one hour from image acquisition to end user

This uses processing from a COTS PC, not the same as the processing the military uses through their custom designed workstations and supercomputers.

Plus, the targets used are very small, less than 20m length, so that adds to the delay.

lol, that show you don't really know how this works.

The problem is how many task you can do with the information.

I can achieve little parity with little processing power, the more processing power only increase the quality of the recognition, but it will not speed up. Because it works on a matrix scale.

Instruction in this case is unchanged, you are looking at the same result (ie I want to track that aircraft carrier) The first thing you do is to local the UI, then compare it with the database, and then generate an outcome.

In the first part, the more UI will only give you more accurate detail, but in this case, let's put an arbitrary number in it, say 6, then if I get 6 UI and compare it, the end result will be the same as if you pick 200 UI and compare it. Just because you are taking 200 UI, that does not mean you cannot do the job with just 6.

The end result is the same, or nearly the same, because you can scale up and down the end result to suit the performance of your computer. That is the reason why the recent publication from China (With 2 fastest supercomputer) still say they are 30-45 minutes from doing ACR in 2017 which is not much of a change from that publication you listed on 2013.

It's like the old question, it take 2 minutes to boil an egg, how long does it take to boil 10 eggs?

 

[gambit]

Most people who design/develop equipment have neither been to the sea or air. People learnt about air and sea after someone designed something that could go out there.

Newton didn't figure out gravity by falling off a building. He observed nature, just like most designers do.

Bad analogy. Newton constantly lived under the laws of gravity. Someone from a landlocked country like Mongolia would have no knowledge of seafaring. If what you say is applicable, we would love to see you read a book on motorcycle racing then challenge Rossi to a race.

So far, NO ONE who has any interests in the DF-21D, have managed to present the sensor capabilities of the weapon. EVERYTHING presented have been speculative. Granted, some are technically based and we should at least grant those speculations reasonable debates. But what have been happening is that those who have that emotional investments in the DF-21D have given the weapon all the benefits of the doubts as to what it carries and can do, while at the same time, place limits on what the American defenses can do.

They say the DF-21D warhead is maneuverable, as in maneuvers like a Top Gun F-14 while at the same time descending at double digits Mach. Not once have they do basic research on how is that possible given the speed to distance coverage. Not once have they presented how the warhead is supposed to maneuver.

 

[randomradio]

Bad analogy. Newton constantly lived under the laws of gravity. Someone from a landlocked country like Mongolia would have no knowledge of seafaring. If what you say is applicable, we would love to see you read a book on motorcycle racing then challenge Rossi to a race.

So most of your scientists and engineers are useless because they are not pilots or seafarers.

Right.

So far, NO ONE who has any interests in the DF-21D, have managed to present the sensor capabilities of the weapon. EVERYTHING presented have been speculative. Granted, some are technically based and we should at least grant those speculations reasonable debates. But what have been happening is that those who have that emotional investments in the DF-21D have given the weapon all the benefits of the doubts as to what it carries and can do, while at the same time, place limits on what the American defenses can do.

They say the DF-21D warhead is maneuverable, as in maneuvers like a Top Gun F-14 while at the same time descending at double digits Mach. Not once have they do basic research on how is that possible given the speed to distance coverage. Not once have they presented how the warhead is supposed to maneuver.

My only major concern has been the missile's warhead, once that is known we will know the capability of the missile. So on this we definitely agree.

 

[gambit]

So most of your scientists and engineers are useless because they are not pilots or seafarers.

Right.

Wrong. The argument is about someone who have never been in the military or worked in the fields directly related to military issues making comments about those fields. The scientists and engineers who created the SR-71 were not biologists working on metallurgy, or geologists working on fuel, or fashion designers working on instruments. Get the picture ?

 

[randomradio]

lol you do know that mean the result is shit, right? I mean, if they actually have a result.

According to the report, the result picked up was accurate enough to determine the object was indeed a cup hidden among clutter.

Then explain to me why when I put a Lancer EVO X in GB Plate
View attachment 393162

It generate a result, but when I put the same car in same color in Australian Plate

View attachment 393163

It then said it cannot be found?

Mind you, the GB result it generate is NOT AT ALL ACCURATE

View attachment 393165

Does this look like a Kia to you??

Just because it said it is independent from reading the license does not mean they don't, just that they say they aren't, otherwise, they would have pick up both car (the same make, model, year and color)

I will let you be the judge

That demo suggests it is purely a license plate reader. If that's the case, then I don't know what they mean by number plate reading.

Anyway--
https://www.idosi.org/wasj/WASJ21(mae)13/19.pdf

And another--
http://www.cs.cmu.edu/~ehsiao/ehsiao_wacv14.pdf
In the future, we plan to incorporate appearance-based features into our view-based 3D models and extend our approach to automatically learn semantic parts of cars that can be highly discriminative for make and model recognition, such as car logos, grill, headlights and taillights.

umm......what about MV Blue Marlin (Beam 63.2 meters)??

https://en.wikipedia.org/wiki/MV_Blue_Marlin

How about Dockwise Vanguard (Beat 79 meters)

https://en.wikipedia.org/wiki/Dockwise_Vanguard

A few cargo ship can have beam over 73 meters, here are the brochure

http://maritime-connector.com/wiki/ship-sizes/

A few ULCC and SLCC crude carrier is actually larger than an Aircraft Carrier with beam over 68 meters

https://en.wikipedia.org/wiki/TI-class_supertanker

View attachment 393169

Do you want me to continue?

Again, just because you said so, does not make it true...

They are all red or blue or some other colour the carrier is not.

Here's an ultra low resolution colour image. Easy to reject when you are looking for a carrier.

and am I hearing this right? (Since there are 12 US Carrier in the world, then we only need 12 person to watch it 24/7??

How about Sat coverage? Are you for sure SAT coverage can cover all 12 carrier 24/7? From my first hand knowledge, US have about 4 times as much Satellite than the Chinese, we cannot have a constant surveillance for a target for over 2 hours in a day, you do know Satellite will go out of range, right? Because Satellite will orbit the earth. Either the earth move you out of the orbit, or your satellite went off the orbit itself. Either way, you can dream of a 24/7 satellite surveillance for a single target, let alone 12 in different corner in the world.

There are brand new satellites that perform earth observation from GEO.

lol, that show you don't really know how this works.

The problem is how many task you can do with the information.

I can achieve little parity with little processing power, the more processing power only increase the quality of the recognition, but it will not speed up. Because it works on a matrix scale.

Instruction in this case is unchanged, you are looking at the same result (ie I want to track that aircraft carrier) The first thing you do is to local the UI, then compare it with the database, and then generate an outcome.

In the first part, the more UI will only give you more accurate detail, but in this case, let's put an arbitrary number in it, say 6, then if I get 6 UI and compare it, the end result will be the same as if you pick 200 UI and compare it. Just because you are taking 200 UI, that does not mean you cannot do the job with just 6.

The end result is the same, or nearly the same, because you can scale up and down the end result to suit the performance of your computer. That is the reason why the recent publication from China (With 2 fastest supercomputer) still say they are 30-45 minutes from doing ACR in 2017 which is not much of a change from that publication you listed on 2013.

It's like the old question, it take 2 minutes to boil an egg, how long does it take to boil 10 eggs?

The issue is you haven't understood the paper. In the paper, they talk about using pixels to detect small objects, but when you are dealing with large objects, you break the image into blocks. When you deal with pixels, you have to deal with a lot of clutter, but when you deal with blocks, you no longer have to deal with clutter. Sea clutter is a non-issue when you are trying to detect something like a carrier. So the computation time reduces exponentially.

This isn't about how many eggs you have to boil, this is about how big the vessel you are using to boil those eggs. If your vessel can only hold one egg, then you are going to have to wait for a long time.

Getting near real time image recognition of such small objects by using pixels is an incredible achievement.

Wrong. The argument is about someone who have never been in the military or worked in the fields directly related to military issues making comments about those fields. The scientists and engineers who created the SR-71 were not biologists working on metallurgy, or geologists working on fuel, or fashion designers working on instruments. Get the picture ?

So you should be clear about that. The way you said it, it was as though you meant only pilots and sailors know about flying and sailing, everybody else can eat grass.

Similarly, you are not a ballistic missile scientist or a commander who uses SRBMs, MRBMs, IRBMs, or ICBMs. So you are not qualified in this field either. Which means whatever information you have is also second hand or less.

Hell, you didn't know that clutter rejection in radars is a non-issue now, and this is from your own field.

 

[jhungary]

According to the report, the result picked up was accurate enough to determine the object was indeed a cup hidden among clutter.

According to the report, that is just a theory, no such thing exist, so how do you know the result is accurate enough to determine anything?

That demo suggests it is purely a license plate reader. If that's the case, then I don't know what they mean by number plate reading.

Anyway--
https://www.idosi.org/wasj/WASJ21(mae)13/19.pdf

And another--
http://www.cs.cmu.edu/~ehsiao/ehsiao_wacv14.pdf
In the future, we plan to incorporate appearance-based features into our view-based 3D models and extend our approach to automatically learn semantic parts of cars that can be highly discriminative for make and model recognition, such as car logos, grill, headlights and taillights.

In the future I can be king of Brunei, I am not interest in talking about things that happen in the future, you can come back to me and say this exist in the future WHEN SUCH THINGS EXIST. but for now, I don't see it.

They are all red or blue or some other colour the carrier is not.

Here's an ultra low resolution colour image. Easy to reject when you are looking for a carrier.

So, first you say there are no ship over 60m beam, now you are saying it does not matter because the side is red and blue??

Anyway, Satellite will not be able to take horizontal or side picture (Then the camera but be of horizontal plane than the ship, which will be in sea surface, not the sky), it MUST be overhead, and deck colour is usually grey or white. Also ,is it possible if indeed US carrier want to blend in civilian shipping, can they paint the deck red and blue?

There are brand new satellites that perform earth observation from GEO.

Do you even know how Geostationary Orbit works??

If it was in GEO, it WON'T MOVE AT ALL (It actual move as the same speed as the earth spinning), that is the reason it was called Geo-STATIONARY Orbit. How can you track something when your satellite does not move?

The issue is you haven't understood the paper. In the paper, they talk about using pixels to detect small objects, but when you are dealing with large objects, you break the image into blocks. When you deal with pixels, you have to deal with a lot of clutter, but when you deal with blocks, you no longer have to deal with clutter. Sea clutter is a non-issue when you are trying to detect something like a carrier. So the computation time reduces exponentially.

This isn't about how many eggs you have to boil, this is about how big the vessel you are using to boil those eggs. If your vessel can only hold one egg, then you are going to have to wait for a long time.

Getting near real time image recognition of such small objects by using pixels is an incredible achievement.

I never said getting NEAR real time in image recognition is not a big achievement, I said it was NOT ENOUGH to track target.

Any computer engineer will tell you that you can clutter your programing in both high and low processing mode. The problem is that, how many subset and how many matrix you can calculate at the same time does not actually directly related to your ultimate goal. the more matrix give you are more detail result, but in this case, you only need meet a minimized requirement to get your result, more processing power is either going to waste or increase the accuracy and definition of the result, but not the time of the result.

Of course, if you want a low processing power computer to computer a high definition result, then it will be slow, however, the speed will be more or less the same if you are using a low processing power to compute a low definition result, it will still be the goal anyway.

That is the reason why the Chinese claim they can do it in 30 to 45 minutes even after computational power increase, actual time increase by less than 15%.

This is what you don't understand.

 

[gambit]

Hell, you didn't know that clutter rejection in radars is a non-issue now, and this is from your own field.

It is...???

https://www.opensesame.com/c/clutte...-moving-target-indication-mti-training-course

Explain to us all how ?

 

[randomradio]

In the future I can be king of Brunei, I am not interest in talking about things that happen in the future, you can come back to me and say this exist in the future WHEN SUCH THINGS EXIST. but for now, I don't see it.

I think you should read the reports first. This 'future' thing they are talking about may already be the past for us today.

The reports talk about recognizing the make and model of a car purely through its shape.

So, first you say there are no ship over 60m beam, now you are saying it does not matter because the side is red and blue??

Anyway, Satellite will not be able to take horizontal or side picture (Then the camera but be of horizontal plane than the ship, which will be in sea surface, not the sky), it MUST be overhead, and deck colour is usually grey or white. Also ,is it possible if indeed US carrier want to blend in civilian shipping, can they paint the deck red and blue?

I have already posted a multispectral satellite image of a cargo ship. The ship's deck is completely red in that case. Yeah, you can change the colour of the carrier's deck as well, but nobody is going to do that. Rather they will use active camouflage systems or similar. Not to mention, you can easily make out the location of the superstructure and other such features on the ship purely through colour. Let's not forget the image is low resolution.

And changing the colour of the deck will only work if you keep the deck empty of aircraft. Or the aircraft themselves will be picked up. I have already presented a paper that explains autonomous detection of aircraft.

You still find it unbelievable but computers have even able to recognize a car using images of a car grill or the shape of its door. The only way you can hide a carrier is by making it homogeneous, and that's impossible.

Here's another paper. Purely concerned with blacking out the license plate and performing image recognition with just the front end design of a car.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.310.6690&rep=rep1&type=pdf

Do you even know how Geostationary Orbit works??

If it was in GEO, it WON'T MOVE AT ALL (It actual move as the same speed as the earth spinning), that is the reason it was called Geo-STATIONARY Orbit. How can you track something when your satellite does not move?

By measuring angle of arrival which is constantly changing because the ship is moving. And when a ship moves from one geographic coordinate to another, within a specific time, you know what's the speed of the ship.

I never said getting NEAR real time in image recognition is not a big achievement, I said it was NOT ENOUGH to track target.

If you detect and identify a target, tracking a target is easy, you have both humans and machines keeping watch. You see, the objective is to make sure you have detected the carrier first, among an ocean full of ships. After that you just maintain a camera lock on the carrier, simple as that.

00:48

The satellite is 'tracking' the aircraft. As it moves from one geographical coordinate to another, you know all the details necessary to maintain track, location, heading and speed.

It is...???

https://www.opensesame.com/c/clutte...-moving-target-indication-mti-training-course

You are confused about what I stated. Water, dust, trees etc are still clutter and are removed.

Explain to us all how ?

If I had the answer I wouldn't tell you, even you understand why this will be a secret. But all I know is the clutter rejection does not remove anything that could be a threat, even birds and insects, so I don't have the answer anyway. And this was stated by that Swedish commander as well, so it holds water.

 

[R Wing]

didn't get your tag for some reason.

There are only one way to sink an aircraft carrier. That is to break into the complex defence system that cover sea, air and information.

An aircraft carrier is protected in 4 tiers defences. The first tier is information.

Information is about threat detection and threat management, it may not be preform by the Carrier Group itself, it can be done by any of the ISTAR platform, ground radar, AWACS, maritime surveillance aircraft, naval surveillance ship, escorts and so on. The first battle is for the attacker to by-pass the detection, and the defender to sniff out attackers, both surface and sub-surface.

The second tier is the defence from the on broad air groups, they can reach out over the reach of the escort ship defences, and they can also detect and attack any threat present in the immediate area, each carrier carry about 50 F-18E/F and in between all those aircraft, they carry 200-400 AAM/ASM as a whole. Couple with AWACS like E-2D and other AWACS/AEWS platform, their job is to seek out the enemy before the enemy can enter the next phase.

The third tier is the defence form escort ship, a CBG would usually travel with 5-6 DDG and 1-2 CG, which mean at any time, the whole 220nm radius will be covered with 700 - 900 cell Mk41 VLS, which can be potential launching a mix of AAM like SM-2 and SM-6, ASM like Harpoon and BMD like SM-3 Travelling with a CBG would also include some Submarine and other assorted ship. Also with 5 inch gun, 3 inch gun and any sort of CWIS, Bushmaster and Machinegun.

When the threat can penetrate all that, then it will the forth tier, which is on broad ship defence. Which will be a mix of sea sparrow, RAM launcher, and CIWS.

That is what it take to sink a Carrier.

Thanks for the info. Is there any (known) weapon and delivery system in the Chinese arsenal that can achieve this?

 

[jhungary]

I think you should read the reports first. This 'future' thing they are talking about may already be the past for us today.

The reports talk about recognizing the make and model of a car purely through its shape.

I have read the report, it is not already past today, as there are no applicable application as is today.

Again, unless you can find an application that support what the report said, I will kindly retract my word.

I have already posted a multispectral satellite image of a cargo ship. The ship's deck is completely red in that case. Yeah, you can change the colour of the carrier's deck as well, but nobody is going to do that. Rather they will use active camouflage systems or similar. Not to mention, you can easily make out the location of the superstructure and other such features on the ship purely through colour. Let's not forget the image is low resolution.

And changing the colour of the deck will only work if you keep the deck empty of aircraft. Or the aircraft themselves will be picked up. I have already presented a paper that explains autonomous detection of aircraft.

You presented NOTHING.

It's all you said what it could do, and provided some theory as to how they do it.

You have yet to provide an actual workable solution to the problem.

I am sorry, but as far as I concern,. you provided jack shit.

I can turn you round and round with a lot of paper about problem and trouble working autonomic recognition, the problem is, you can dig around these paper for theory, but at the end of a day, you need a working prototype to show people you can do it. And it's not you can do it "In Theory" or you can do it "In the future"

You still find it unbelievable but computers have even able to recognize a car using images of a car grill or the shape of its door. The only way you can hide a carrier is by making it homogeneous, and that's impossible.

Here's another paper. Purely concerned with blacking out the license plate and performing image recognition with just the front end design of a car.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.310.6690&rep=rep1&type=pdf

Another theory.

You said the previous one were independent from license plate, yet, I have show you they aren't. How do you know this one is not following the same way?

Not to mention that problem itself have conceded they cannot get the make and model every time.

By measuring angle of arrival which is constantly changing because the ship is moving. And when a ship moves from one geographic coordinate to another, within a specific time, you know what's the speed of the ship.

How?

First thing first, GEO Orbit does not move, how are you going to get the target location if it move out of the GEO SAT boundary?

Second, ship can change course and speed, you cannot know what the speed is, maybe with refer to the Satellite, not with reference to the ground.

If you detect and identify a target, tracking a target is easy, you have both humans and machines keeping watch. You see, the objective is to make sure you have detected the carrier first, among an ocean full of ships. After that you just maintain a camera lock on the carrier, simple as that.

00:48

The satellite is 'tracking' the aircraft. As it moves from one geographical coordinate to another, you know all the details necessary to maintain track, location, heading and speed.

Dude, we are not talking about tracking, we are talking about Automatic Tracking.

How do you know if the satellite does not have manual input behind the tracking?

You keep saying "If", "In the future", "Maybe" the problem is, you don't know what they can do.

If I can get a guy to stick his head out, I can shoot him in the head very easily, that does not equal to it's easy to shoot people in the head.

You failed to understand the basic principal of ISTAR, which is "If" you have a target, of course everything is smooth sail from then on, the problem we have been discussing is HOW YOU CAN ACQUIRE A TARGET ON ITS OWN IN THE FIRST PLACE?

Thanks for the info. Is there any (known) weapon and delivery system in the Chinese arsenal that can achieve this?

Currently, there are weapon that Chinese process can penetrate one of the defence, but there does not exist a weapon that can penetrate all of them and guarantee a hit.

 

[randomradio]

Not to mention that problem itself have conceded they cannot get the make and model every time.

Yes, because detecting a car is extremely difficult. There are so many different designs of cars, but many of them look very similar at the same time. But a carrier, not so much.

If a computer can recognize a car just by looking at its grill, then detecting a carrier with a massive superstructure, carrying aircraft that puff out huge amounts of heat is not difficult. You can easily distinguish a carrier from a regular ship.

You said the previous one were independent from license plate, yet, I have show you they aren't. How do you know this one is not following the same way?

It's simple. They actually say that they are not using license plates.

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.310.6690&rep=rep1&type=pdf
Automatic vehicle surveillance is an increasingly important technology due to a rising trend in license plate cloning. In the UK, for example, tens of thousands of license plates are cloned every year. Reasons for plate cloning range from evading speed camera fines and congestion fees to selling stolen cars and even disguising a vehicle for use in a serious crime. To deal with this, automatic recognition of the actual vehicle itself is required to supplement standard automatic number plate recognition (ANPR). Vehicle recognition systems usually either classify the vehicle into generic classes (car/lorry etc) [3], or they classify the vehicle into specific ‘make and model’ classes. Here we focus on the latter application of make and model recognition (MMR).

https://www.idosi.org/wasj/WASJ21(mae)13/19.pdf
Several computer systems were proposed to identify cars via License Plate external features. Recognition which is low cost because they do not need any expensive hardware for high quality video or other sensors. However, they fail when an automobile has fake or no plates and are able to avoid recognition. Vision based car Make and Model Recognition (MMR) can overcome such cases and provides additional security to License Plate Recognition systems.

http://www.cs.cmu.edu/~ehsiao/ehsiao_wacv14.pdf
In this paper, we have developed an approach for verifying the make and model of a car from a single image taken from an arbitrary viewpoint using view-based 3D curve models. Our experiments show that these models can be accurately aligned to test images after which chamfer distance and orientation distance can be used to recognize the car make and model under wide range of viewpoints.

You see, if cars can be recognized without number plates, then what's stopping ship recognition, it's even more easy.

First thing first, GEO Orbit does not move, how are you going to get the target location if it move out of the GEO SAT boundary?

That's 1/3rd of the earth's surface.

Anyway--
http://www.popsci.com/gaofen-4-worl...ontinues-chinas-great-leap-forward-into-space
In the Gaofen 4's case, its range of view is a 7,000km by 7,000km box of 49 million square kilometers of Asian land and water in and around China.

You only need a handful of satellites to watch the entire globe, even if that takes forever.

Second, ship can change course and speed, you cannot know what the speed is, maybe with refer to the Satellite, not with reference to the ground.

You can easily infer the ship's course by observing the wake. Similar to the airplane in the video.

Dude, we are not talking about tracking, we are talking about Automatic Tracking.

How do you know if the satellite does not have manual input behind the tracking?

You keep saying "If", "In the future", "Maybe" the problem is, you don't know what they can do.

If I can get a guy to stick his head out, I can shoot him in the head very easily, that does not equal to it's easy to shoot people in the head.

You failed to understand the basic principal of ISTAR, which is "If" you have a target, of course everything is smooth sail from then on, the problem we have been discussing is HOW YOU CAN ACQUIRE A TARGET ON ITS OWN IN THE FIRST PLACE?

Although a satellite can track an object automatically for a short duration, until a positive identification is made and the human operator is informed, why is there a need to rely on the satellite alone? I already told you, once a carrier is detected, you can have a guy watching it the entire time, which you can repeat for the whole fleet.

The very point of building closed bases for submarines is you don't want anybody watching when the sub enters or leaves the base. There is no such cover for carriers. In fact, if you have the resources, you can permanently assign a satellite or two for carrier surveillance from the day its keel is laid to the day it is keelhauled.

 

[Gomig-21]

So far, NO ONE who has any interests in the DF-21D, have managed to present the sensor capabilities of the weapon. EVERYTHING presented have been speculative. Granted, some are technically based and we should at least grant those speculations reasonable debates. But what have been happening is that those who have that emotional investments in the DF-21D have given the weapon all the benefits of the doubts as to what it carries and can do, while at the same time, place limits on what the American defenses can do.

The only test we know of so far of the DF-21D (at least the one I'm only aware of) or it's near predecessor was on a still platform target on land with two hits. The interesting thing about that is this could have been easily done with subsonic cruise missiles. Not that there is doubt on the test, just that it doesn't mean as much as some might think. Once you get into the area of ballistic missile operations, it's a completely different story. To be able to successfully narrow in on moving targets in open sea where you have to account for the effectiveness of the missile's sensor capabilities, target acquisition and seeker capabilities and final homing at double digit mach speed is quite different. That elevates the degree of complications for a successful hit. None of that was tested on that specific test AFAIK.

Evasion of countermeasures was most likely never considered. If its terminal velocity is in fact somewhere between mach 8 & 10, THAAD and other airborne or SAM interceptions would probably not be useful with high probability misses at that incoming speed. The only real and effective countermeasure against something like the DF-21D is to jam any associated "over the horizon" radars, spoof them with decoys or smoke, although tough to smoke off an entire carrier but it's possible with several surrounding ships. The missile's mid-course updates (if it truly has those capabilities which we don't know) can be jammed since it will need to be constantly fed new coordinates on a moving carrier and even if it got through all that and locked on to target, its seeker can be electronically jammed or spoofed with decoys. It's a tall order to hit those decks with what the US has for a defensive apparatus that consists of all these layers that don't even include any actual interceptions. What about countermeasures we don't even know about? lol. There is something called 'secrecy' that even the US employs so no one else can know about them and develop a way to circumvent them.

The other thing is, for a saturation attack to work, they all have to get past the above mentioned countermeasures in the first place, otherwise you'll be destroying a lot of the empty and beautiful, South China Sea with its many plump tuna, dolphins, swordfish and octopus. It would be a shame if that happened.

 

[gambit]

If I had the answer I wouldn't tell you, even you understand why this will be a secret. But all I know is the clutter rejection does not remove anything that could be a threat, even birds and insects, so I don't have the answer anyway. And this was stated by that Swedish commander as well, so it holds water.

You do not have any answer.

http://www.militaryaerospace.com/articles/pt/2017/01/wind-farm-radar-clutter.html

For the above situation, what happened was that windmills created anomalous readings for the local weather radar, in other words, the windmills were clutter that interfered with the normal operations of the weather radars. The windmills were modified to be 'stealthy' but even so, their signatures had to be measurable so that the weather radar can discount them in its data processing.

When an attack aircraft flies over man made terrain, aka 'cities', this kind of clutter can -- not always -- create unanticipated interference that no radar can compensate for. So for you to say that clutter rejection is no longer an issue means you do not know what the hell you are talking about, especially when clutter rejection algorithm is STILL being taught.

This is what separate me from the rest on this forum in that I do not seek to discredit just because I do not like a specific country. If someone make a claim or post a source that is technically dubious, then I will challenge and I will support my arguments. In your case, you are no different than many in that you just seek to discredit the US simply because you have some kind of animosity against US.