The Military-Industrial Scientific Research System of the Academic Master

Chapter 959 Target Contour Recognition

Nowadays, if Chang Haonan says that there is no problem in solving partial differential equations, then at least in China, it is unlikely that anyone will easily jump out to question this conclusion.

Therefore, everyone basically agreed that their next step was to find the correspondence between image processing methods and specific partial differential equations.

However, for a real ocean surveillance system...or for the entire image recognition and processing technology, this is still only the first step.

Even the first action in the first step——

Even as far as the "image segmentation" link is concerned, there are still other needs that need to be solved.

Of course, the ideas given by Chang Haonan are indeed novel enough.

So soon someone started to draw inferences:

“Mr. Chang, if a single level set function can divide the image into the background area and the target area, that is, a two-phase segmentation, then if we use multiple level set functions at the same time, can we complete the multi-phase image segmentation? segmentation?"

"In theory, of course."

Chang Haonan replied:

“When an image contains a lot of information and cannot be simply divided into two parts: target and background, it is necessary to consider the multi-phase variation level set problem.”

"However, the complexity of polyphase problems goes to a higher level."

After he finished speaking, he turned back and drew a square on the blackboard.

"We can simply think that each level set equation delimits an area in the image, which is this square."

"When there is only one square, the image will be, and will only be, divided into two parts: inner and outer. There is no second division."

"but."

Chang Haonan drew a second square on the blackboard and marked 1 and 2 respectively on the two figures:

"When the second square appears, there is not one more variable, but two more variables, that is, the second square itself, and the positional relationship between the two squares."

"If the two squares do not overlap, then the entire image will be divided into three areas, namely inside No. 1, inside No. 2, and outside No. 1 and No. 2."

"And if two squares overlap, then the entire image will be divided into four areas, No. 1 inside and No. 2 outside, No. 1 outside and No. 2 inside, No. 1 inside and No. 2 inside, and No. 1 outside and No. 2 outside..."

“Actually, when we set multiple level set functions to divide an image, the image can be divided into 2^N areas at most, and N areas at least, so the difficulty of classification is not the same. It’s not a straight rise, it’s an exponential rise…”

"..."

If this part of the content is described in mathematical language, it is actually quite complicated, but Chang Haonan explains it completely in vernacular, which makes it easy to understand.

"so……"

The person who just asked the question also understood:

"We still need to specify a level set design plan before we can specifically complete the regional division of the polyphase diagram?"

"That's right, but it's not complete either."

Chang Haonan threw the last small piece of chalk into the chalk box next to him:

“It seems like there is just one more degree of freedom in the software design logic, but in fact, in addition to the extreme case of dividing it into N or 2^N areas, additional constraints must be added, and the amount of calculation is always relatively large. frightening."

"Even the simplest polyphase image segmentation, which is the use of two level set functions to divide 4 regions just mentioned, will also produce 4 local grayscale fitting functions, and then use the variational level set method to minimize the total energy spread Function, this process requires solving 8 Euler-Lagrange equations, you can figure out more complex situations by yourself..."

"In addition, the probability of capturing two high-value specific targets in one photo is very low. Even if this happens by chance, as long as any one of them can be identified and the image is correctly classified as a key point and sent back to On the ground, the effect is the same. After all, we still have ground analysts and military commanders who don’t just rely on the information provided by satellites to fight.”

"..."

While Chang Haonan was speaking, he briefly wrote on the blackboard the solution results of the fitting function and the average gray value in the classic case.

Obviously, unless the situation is very coincidental, such a process must not be suitable for completion on a spaceborne computer, at least within the next few years.

"So, this aspect will not be the focus of our research for the time being?"

Lin Senming put down the pen in his hand and moved his wrist slightly and asked.

"Not necessarily."

Chang Haonan wiped his hands and sat back at the head of the conference table:

"If the stability of the equation is relatively good, then the semi-implicit difference scheme can actually be used to solve it numerically. Except that it takes a long time, there is actually no other obstacle..."

"In short, I have ready-made theoretical ideas and calculation tools. If Mr. Lin has enough funds, computing power and manpower, I can ask Teacher Yao Mengna from our laboratory to be responsible for connecting with you in detail."

"Although the satellite terminal cannot do this, there is still some value in post-processing the returned images on the ground terminal."

"In addition, the field of view can be broadened. This type of technology does not necessarily have to be used on satellites. Aircraft, ships and even vehicle platforms can support much more computing power than satellites. They can be used in the future. Benefiting from it, this does not include the potential in the civilian field..."

Although Lin Senming was one of Chang Haonan's deputies in the Ocean 1 satellite project, the former was from the Aerospace Science and Technology Group after all, and there was actually no organizational relationship between the two parties.

Therefore, Chang Haonan could only use the method of drawing cakes to induce a wave of temptation.

The main reason is that his own energy and resources at hand are always limited.

It is better to take advantage of this opportunity to give full play to the ability of the non-staff deputy director of the Science and Technology Commission and mobilize the subjective initiative of the units at the same level.

In this way, when building a strategic reconnaissance aircraft or other similar needs arise in the future, you will have a ready-to-use shelf product.

The research and development cycle can be greatly shortened.

"This... I will report this to my superiors when I get back."

Since it is not directly related to Ocean One, it is naturally impossible for Lin Senming to admit it right away.

But it was not difficult to guess from the almost glaring look in his eyes that he was actually quite interested in it.

As for the expression "reporting to superiors".

It's just a classic tactic of delaying the attack.

"Comrades, if there are no other questions."

Chang Haonan said as he opened his laptop and connected it to the projector on the side:

"Then next, I will introduce the mathematical principles of the variational level set model in detail from a theoretical level..."

There are basically no contestants who can sit in this conference room who have no idea about the field of image recognition and processing.

Therefore, Chang Haonan can skip the introduction of several common image segmentation models to a certain extent.

However, he was still unable to finish all the content he had planned in advance on the same day.

The two main contents are the curve evolution theory and the variational level set method, which are indeed somewhat difficult to understand.

Especially for some technicians with engineering backgrounds, it is even more abstract to the extreme.

Fortunately, the most important person, Lin Senming, majored in applied mathematics, but he followed Chang Haonan's ideas throughout the process.

Even so, this meeting... or rather a series of meetings, lasted for four full days and did not finally end until Friday.

And just this week, another thing closely related to Chang Haonan was happening simultaneously without him knowing it at all...

Chapter 960/1340
71.64%
The Military-Industrial Scientific Research System of the Academic MasterCh.960/1340 [71.64%]