Chapter 958 How to identify an aircraft carrier?
It is naturally impossible for the delegation from Snecma to wait in Shengjing for ten days.
As for the idea of further cooperation, although there is one, it is impossible to decide on the spot.
Therefore, after agreeing on some details the next day, Ian Vista and other core members of the technical team returned to France.
Only a few people were left, together with Bernard Sean and the others, waiting for China to officially deliver the batch of samples for testing.
Liu Yongquan, on the other hand, reported the meeting status and of course the new production requirements to the group level as quickly as possible——
The technical team of the Haojing Institute of Optics and Mechanics has successfully assembled the first one at Liming Factory. An ultra-short laser processing equipment used for mass production, but it is still in the process of debugging and testing and will not have the ability to be put into actual production before the end of the year.
Therefore, at least for now, this batch of M88-3 parts still have to be handed over to the machine in the Torch Laboratory for hole making operations.
However, since the technology development phase has been basically completed, including the fact that even the parts of the M88-3 are not produced for the first time, Chang Haonan only needs to handle it.
The specific work at the software and hardware levels can be completed by Li Yabo and Academician Hou.
Of course, this is not because Chang Haonan deliberately wanted to be a hands-off shopkeeper.
Although he is no longer interested in the M88-3 project, the SeA series engine, as the first medium or higher bypass ratio engine launched by the Aviation Industry Group, is still very worthy of attention.
In fact, Chang Haonan even planned to use the development of SeA650 as a training and production before the official establishment of the WS20 project, so he specially invited many members from the Haofa Group to participate.
The reason why he had no time to pay attention to the situation on the aviation engine side was because after he came back from Haojing after watching Figure 160 last Friday, he only had time to rest for one day before he received a call from the Aerospace Science and Technology Group. Side phone.
Invite him to a project development meeting.
As for the content of the meeting, it was naturally about the onboard software of the Haiyang-1 satellite.
It was also a coincidence that Chang Haonan happened to have some new ideas in target contour recognition while developing new algorithms for TORCH Multiphysics software in the past few months.
Therefore, he was preparing for this matter seriously for a period of time.
It was almost at the same time that Vesta and others boarded the plane to return to France.
Chang Haonan is in his laboratory, giving a speech to a group of technicians from the Institute of Space Technology, the Fifth Aerospace Institute...
lecture.
That’s right.
Similar to most of the projects he has participated in in the past.
During the project development meeting, Chang Haonan introduced his contour target extraction model in the shortest possible length.
But in an environment like a conference room, it was obviously not effective enough for him to speak alone.
Besides, there were also experts in rocket technology who were in the pure hardware field and had nothing to do with satellites participating in that conference. It was not easy for him to bring everyone together to listen to his long speech.
Therefore, after Chang Haonan realized the problem, he immediately chose to give up.
Then find some people related to software development and software and hardware adaptation, and hold a small meeting.
"Mr. Chang, the parametric active contour model (snake model) you mentioned yesterday has been systematically demonstrated in our institute in the past two years."
The person who spoke was sitting in the chair. There is a young technician at the head of the left side of the conference table named Lin Shenming:
"But in the end, it was found that although the snake model can provide smooth and closed edge contours that meet pixel requirements, it cannot adaptively control changes in the curve topology, and it is difficult to clearly segment the concave areas of the target object, which is very difficult for ships... Uh...it’s a big problem for ocean feature target recognition.”
Although he is only in his thirties, Lin Shenming is actually the deputy chief designer of the Haiyang-1 satellite ground application system.
If it weren’t for Chang Haonan who was standing at the front at this time, he would obviously be one of the most dazzling young talents in the entire Ocean One project.
Of course, his current problem does not involve any personal emotions.
I just have some concerns about the technical route of the active contour model.
Although Ocean 1 has solved the hardware problems and can even be said to have acquired the world's first-class optical system, it is still active in low-Earth orbit at an altitude of 400 kilometers, and the photos taken are still blurry after being enlarged.
Therefore, it is extremely important to accurately identify specific contours from a bunch of vague chaotic targets.
The prerequisite for identifying the outline is to segment the outline shape from a blue or black background.
If Haiyang-1 is just like the name, it is a conventional marine resource satellite, then it is not a big deal that the concave area cannot be divided, that is, the area with an internal angle greater than 180°.
But as a de facto ocean surveillance verification star, this problem is completely unacceptable. Aircraft carriers are huge, that’s true.
But in the ranks of civilian ships, there is no shortage of giants that are close to or even exceed the size of aircraft carriers.
The biggest feature of the aircraft carrier - the all-access deck, from the perspective of satellite observation, is similar to the vast front deck of some civilian transport ships.
Even a container ship filled with containers can easily be misjudged.
In this way, there are only two features that can be easily exploited.
First, move to the ship island on one side.
Except for warships such as aircraft carriers and amphibious assault ships that require aircraft carriers, no other ships will use this bizarre design.
Civil ships will set the cab as close to the front or rear end of the hull as possible.
But the problem is that the ship island is indeed a bit too small.
In particular, nuclear-powered aircraft carriers do not need flues, so they can make the island very small and very far back.
And it also highly overlaps with the deck part.
Whether it is radar, visible light or infrared light, it is difficult to identify it if the resolution is insufficient.
In contrast, another feature, the irregular hull profile formed by the combination of the angled deck and the take-off deck, is more friendly.
Due to floor area ratio considerations, it is impossible for civilian ships to design such a zhilengbaqiao shape.
Even small and medium-sized aircraft carriers without angled decks and amphibious assault ships can be excluded to a certain extent.
And more importantly, an outline of 300 meters long and 40 meters wide is still barely segmentable with current satellite resolution.
So, it is necessary to solve the problem of recognition rate of sunken areas.
Otherwise, if we just divide the angled deck into rounded arcs, it would be almost like an oil tanker, with no features left at all.
"Mr. Lin is right."
Chang Haonan strolled to the blackboard:
"It is difficult for several traditional active contour models to be directly applied to the image recognition algorithm of our Ocean One."
"But when I was writing a paper before, I saw it accidentally One idea is to use the variational level set method to improve the active contour model. "
"So far, the main product of this idea is the geometric active contour model, and further, the use of Mumford-Shah functional. Geodesic active contour model developed by segmenting the variational level set method. ”
“Of course, since the level set method has had non-conservation problems in the past, these two methods are facing curves. The adaptability when the topology changes is still limited. "
"However, if you have read my previous article in the first issue of JCAS, you will know that I have solved this problem from a theoretical level. One question..."
After finishing speaking, Chang Haonan wrote the first equation on the blackboard:
p(F)=e^(-βE|F|)/Z.
And in this At that moment, another person below raised his hand:
"Mr. Chang, I have read most of the text in the first issue of JCAS, but if I remember correctly, the article you published It seems to be used for multi-phase flow simulation?”
“Yes”
Chang Haonan happened to have finished writing the equation at this time, so he turned around and replied:
“But things at the level of mathematical principles remain the same.”
“The difficulty of multi-phase flow simulation , also because the topology of the phase interface is highly uncertain, so the moving interface needs to be described as a time-varying level set."
"Therefore, this method can be used with almost no modifications. Application in the curve evolution of partial differential equations"
After finishing speaking, he pointed to the blackboard again:
"This is the Bayesian form corresponding to Gibbs' formula in statistical mechanics, and the variational method I just mentioned happens to be exactly consistent with it in form..."
"And The image processing method based on partial differential equations is essentially based on the continuous mathematical model of the image, assuming that the image changes according to a specified partial differential equation, and the solution of PDE is the desired processing result. ”
< br>"As for how to determine the form of this formulated partial differential equation..."
Chang Haonan said while drawing an arrow between "image change" and "partial differential equation":
< br>"Generally speaking, the desired image change is compared with some mathematical physical process, for example, the smoothing of the image is analogized to the diffusion process of impurities. Of course, most processing methods will not be so simple, and this is not the case. This is exactly what we need to study at this stage.”
"As far as my current research progress is concerned, as long as I can get to this point."
He circled the five words "partial differential equations" again:
"The following The numerical solution process is almost not a problem! ”
(End of this chapter)