1995, Starting as a Criminal Investigation Expert

Chen Yan was deeply impressed by Li Shu, but had limited contact with him.

Even though we've been working in the same department every day for the past few days and have had a lot of contact, our roles are still different.

She was just a "student" pursuing further studies, while Li Shu, who was a year younger than her, had already become an associate researcher and deputy director.

Therefore, in their daily interactions, due to the difference in their status, Chen Yan always felt that Li Shu had a natural air of superiority.

Even if Li Shu doesn't put on airs.

But right now, seeing Li Shu acting like a "scoundrel" and arguing with Wei Ranhui over a $10,000 limit, and even massaging his shoulders to curry favor with him, is quite shocking.

Surprisingly, Li Shu seemed much more vibrant.

Of course, the $10,000 limit also surprised her quite a bit, and she wondered what Li Shu was going to do with it.

Li Shu paused.

Wei Ranhui removed Li Shu's hand and said with a smile, "How about you teach us this algorithm, and I'll see how I can help you figure out a solution."

Li Shu sat back down.

"Director Wei, please don't say that. You know I'm often away on business trips and don't spend much time at the institute. Otherwise, I would have told you about that algorithm long ago."

The reason for going directly to Wei Ranhui instead of Liu Yaoqing was mainly because they didn't want to put Liu Yaoqing in a difficult position.

In a institute with so many departments, there are needs. If you use up so much at once, then others won't have any. As the director, Liu Yaoqing has to treat everyone fairly. There's no reason to make things difficult for him.

It's up to the subordinates to find Wei Ranhui themselves; it depends on their personal connections and abilities.

So the fear is that Wei Ranhui won't make any demands.

Wei Ranhui immediately laughed: "So, it's because I was thinking about you working at the institute recently that I humbly asked for your advice."

Li Shudao said, "Oh, what do you mean by asking for advice or exchanging ideas? We're exchanging ideas."

Chen Yan, standing behind, saw Li Shu's reaction.

She had excellent grades since childhood, the kind of child that other people admired. She went to a good university and went to work in the provincial capital after graduation. She was young and didn't really like the kind of "traditional craft" units in the province where people had to greet and fawn over each other.

I even felt a bit contemptuous from the bottom of my heart, believing that people should rely on their real skills.

But at this moment, seeing Li Shu "flattering" Wei Ranhui, I didn't feel contempt. I just felt that Li Shu was genuine, understood human relationships, and had real skills.

If Wei Ranhui didn't understand this algorithm, would he have given up on it?

Learning on the spot, Wei Ranhui immediately went out and gathered the people in his department to prepare for the lesson.

Li Shu took a notebook and pen from Wei Ranhui's desk, handed them to Chen Yan, and said, "Take some notes in a bit, and study them later. There are more and more CCTV cameras being installed now, and I have a feeling that they will become an important tool for solving cases in the future."

:

Chen Yan made a note of it, then asked in a low voice, "Director, what is this 'AL' you mentioned? Why does it cost ten thousand US dollars?"

Li Shu asked in surprise, "You don't know?"

The Criminal Police Academy should have AI.

But then she thought, this stuff is in foreign exchange, it's ridiculously expensive, how could an ordinary student like Chen Yan possibly have access to it?

Li Shu explained, "It's a retrieval system that can search for cutting-edge international journal articles."

After explaining the situation, Li Shu continued, "Chen Yan, I'm preparing to work on a big project recently, which requires me to review a lot of information. Over the next period of time, I will arrange for Yan Bei and his team to conduct research in various places. The results of their research and the information they collect will be given to you to organize."

It's too much of a hassle for her, a young woman, to run around all day. Li Shu plans to have Chen Yan take charge of the data organization.

Chen Yan nodded: "Director, don't worry, I will definitely do my job well."

You can't not work hard.

Looking up information requires foreign exchange; if you're not careful, you could end up losing a lot of money.

Upon arriving at the department's conference room, where Wei Ranhui and his colleagues usually exchanged technical information, they found themselves in the meeting room.

Li Shu was greeted with applause as soon as he entered.

"You're too kind. Let's exchange ideas," Li Shu said politely, then asked, "Director Wei, are you facing some difficulties?"

It's likely that a case came up that the original technology couldn't solve, which is why they thought of the new algorithm they mentioned.

Wei Ranhui sighed softly and said, "You're someone who travels frequently for business and cases, so you guessed right away. There is indeed a case. Why don't you take a look at it for me in a bit?"

Li Shu smiled and said okay, then began to explain the algorithm.

"We used to use inverse filtering, but we couldn't accurately determine the point function diffusion, so no one used this algorithm, which is why Wiener filtering gradually emerged."

These words were actually meant for Chen Yan.

The earliest image processing algorithm was the inverse filtering algorithm, which was from the 1960s and 70s and had major flaws. That's why the Wiener filtering algorithm was developed, which is now the mainstream algorithm and the main technique used by Wei Ranhui and his team.

"However, Wiener filtering can only process single-frame images, and the pf function is entirely estimated manually, which is too dependent on personal experience, especially when there is a lot of noise, so the recovery ability is limited."

Everyone nodded in agreement, and Chen Yan, sitting in the corner, was also taking notes diligently.

Li Shu glanced at her, slowed his speech, and waited for her to finish taking notes.

After Chen Yan finished writing, she looked up at Li Shu. Their eyes met, and she immediately realized that Li Shu was waiting for her.

A slight warmth touched his heart, and he thought to himself that Li Shu was truly thoughtful.

Li Shu withdrew his gaze, finished explaining the basics, turned to write formulas on the blackboard, and said, "The algorithm I'm talking about has a core mathematical principle of imaging degradation model."

:

Mathematics is just the theoretical foundation, and it would be too troublesome to explain in detail, so Li Shu will just talk about the key points and let Wei Ranhui and the others study it in depth later.

"The process involves two steps. The first step is pixel registration, which has two approaches: phase correlation and pyramid iteration. The only problem is that our current recording hardware is not up to standard, with only one frame per second, so the registration algorithm is useless."

The video footage obtained in the Yuzhou case was not restored to its original clarity.

Instead, they used the traditional Wiener filtering algorithm to process each frame, and then inferred the outline of Lü Jun's facial features by observing changes in light and shadow.

This algorithm hasn't been used yet.

Now that we're getting down to business, all ten people present, including Wei Ranhui, started taking notes.

There were also cameras on the frame.

Li Shu would still record videos whenever he had free time and then send copies to the grassroots level.

I was in Hong Kong for over a month recently and haven't had the time.

Li Shu continued, "After registration, the image needs to be reconstructed. As for algorithms, I know of four types."

Based on the system documentation, the most classic algorithm is BP, which is iterative directional projection.

However, given the current hardware conditions of one frame per second, this algorithm has significant flaws, and iteration can easily amplify noise.

It's safe to say that it's almost never used.

Theoretically, the best algorithm is the fourth one, map, which is the maximum a posteriori probability algorithm.

"This algorithm has great advantages, but the problem is that the computational load is too large. If you really want to process a single frame of image, even with our latest computer, I estimate it will take half an hour."

Most importantly, knowing the algorithm is useless.

This is just an algorithm; it needs to be written into software. Li Shu doesn't have that ability; his computer skills aren't good enough.

"So, Director Wei, how exactly to implement and use it is up to you to figure out."

Wei Ranhui and his team are the real professionals.

Wei Ranhui stared at Li Shu, then suddenly stood up and pulled Li Shu outside the door.

"My God, Li Shu, are you serious?"

As soon as the door closed, Wei Ranhui exclaimed in surprise.

When Wei Ranhui first heard Li Shu mention a new algorithm, but Li Shu didn't go into detail, he thought it might be research based on Wiener filtering.

When Li Shu finished speaking, he was dumbfounded.

:

While the basic imaging degradation model was a concept from the 1980s, the subsequent pixel registration and image reconstruction algorithms were concepts he had never heard of before.

Even more outrageous is that Li Shu mentioned four algorithms at once, which is clearly an attitude of not stopping until he makes a shocking statement.

Wei Ranhui suddenly remembered something and hurriedly asked, "Li Shu, have you published any papers on these algorithms?"

He felt that any one of these studies could lead to countless papers, making them worthy of in-depth research.

If we calculate it this way, Li Shu's casual remark, without exaggeration, has helped the Computer Applications Department solve at least two years' worth of research tasks.

This is a huge favor.

"I don't have the time for that."

Li Shu shook his head. Image processing has a very small application scope now, and he really doesn't have the energy to study it in depth.

"And Director Wei, don't even think about publishing papers. The things I mentioned might just be things you haven't explored yet, but they've actually been around for the past few years."

According to the literature, BP was proposed in 1991, and MAP was a concept that emerged in the mid-1990s.

It's been happening in the last few years.

However, these algorithmic theories are not applied in the field of criminal investigation.

Sun Bomin was right when he said that criminal investigation was a minor matter.

These algorithmic theories are concentrated in the fields of astronomical observation, satellite remote sensing, military reconnaissance, and medical imaging.

Compared to these things, criminal case investigation is really nothing.

So Wei Ranhui hadn't been exposed to it and thought it was a new concept, which is why he reacted so strongly.

However, Li Shu had a sudden thought and changed his mind, saying, "But if you really want to write a paper, it's not impossible. For example, you can find some information about maps and study them in depth."

Based on the available information, the most balanced and suitable algorithm for the future is map.

Now that this concept has just been proposed, others are researching it, and Wei Ranhui and his team can also research it and write papers.

Even the backpropagation (BP) algorithm mentioned earlier can still be studied and used to write papers.

These are all cutting-edge theories from abroad, which are probably only just being explored or just beginning in China.

"I really don't know."

Wei Ranhui was taken aback; he hadn't really paid attention to these studies.

I couldn't help but feel a pang of regret, realizing that while I was in charge of the institute's research funds, I was neglecting other departments.

Now it seems I'll have to spend money to look up some cutting-edge information myself.

:

However, regarding the paper, Wei Ranhui said, "Goodbye, Li Shu, we won't write the paper. This is your research, so you should write it yourself."

If I were to use Li Shu's idea to write a paper, wouldn't that be plagiarism?

What a huge advantage this is!

Li Shudao said, "How about this, Director Wei, let's collaborate and write it together. Once you've actually written the paper and published it, I can just be listed as an author."

These algorithmic theories, if studied in depth, involve mathematics and computer science, and even Li Shu would have a hard time mastering them.

The best option would be to propose a concept yourself, let Wei Ranhui and his team complete the research, and then simply sign the name.

Wei Ranhui clearly wanted to, and with an awkward smile, said, "Then we're really taking advantage of you."

He really wanted it.

Of all the departments in the institute, the Computer Applications Department is the one that gets no attention.

To put it bluntly, it's because they didn't achieve any results.

Do some new research, write some cutting-edge papers, and let other departments see what we've got.

"By the way, Li Shu."

Wei Ranhui leaned closer and lowered his voice, saying, "Go check the data. Is ten thousand US dollars enough? Let me tell you, we only have twenty-seven thousand dollars left in funding this year. See if that's enough."

"If that's still not enough, I can help you find a way from other institutes and look into it for you as well. Besides, if we're going to collaborate on a paper later, we'll need to look up information too. It'll be more convenient if we split the accounting into two departments."

Li Shu immediately smiled and thanked him.

Back in the conference room, Wei Ranhui asked everyone to listen attentively.

Li Shu didn't hold back; he explained all four algorithms and two registration models in the way he could understand them.

They're not stupid; they can tell the value of this information.

After Li Shu finished speaking, thunderous applause erupted.

Chen Yan also put down her pen and clapped along, staring intently at Li Shu with admiration in her eyes, feeling as if Li Shu was shining.

Li Shu waved his hand, telling them to go ahead with their work, and pulled Wei Ranhui aside to ask what the case was.

To be honest, I'm more interested in the case than in the research.

: