Chapter 125: Upper Limit
Forensic image analysis is a criminal science field with an extremely high upper limit and a very low lower limit.
In terms of upper limit, there are already countless noise and clear-image models for image denoising alone, and the number keeps growing.
Different noise models and clear-image models, combined with various solution methods, give rise to a wide variety of denoising algorithms—numbering in the hundreds.
For example, the well-known 3D denoising, NL-Means denoising, BM3D denoising, and others.
But writing algorithms and building models is not the highest upper limit of forensic image analysis. To practically handle the images and videos encountered in this field, one must not only master various algorithmic models but also possess substantial experience with workflows and methods, and solid command of related technologies.
Of course, such an upper-limit requirement cannot be met even nationwide, let alone within a single province.
In places like Ningtai County, image offices typically exist only to meet the bare minimum: locating surveillance footage and identifying suspects.
Sometimes even that fails, and it doesn't matter.
Zhuang Wei and the others originally worked only to meet the bare minimum. Now Jiang Yuan has shown them what a near-upper-limit capability looks like.
Crossing this chasm would inevitably be extremely painful.
Seeing it but being unable to touch it is truly frustrating.
Jiang Yuan pulled up another video, glanced at the third-generation image technician beside him, and said simply: "Different images require different methods. With current technology, no single algorithm can handle all image types. You can try a few common modules, but the results won't be good."
As he spoke, Jiang Yuan pulled up wavelet transform—a technique entirely different from the previous one—and began filling in parameters.
The core of image enhancement is the algorithm; using different software is essentially choosing different algorithms.
For the user, after selecting an algorithm, the most important task is filling in parameters—and there are many of them.
Surveillance videos from different cases vary greatly, and each requires a distinct processing workflow—problems that are easy for humans to solve but difficult for programs.
For instance, the video Jiang Yuan was now viewing was a long-distance backlit surveillance clip. The two targets appeared in the lower-left corner, slightly distorted.
These are precisely the areas algorithms struggle to address, requiring Jiang Yuan to manually set parameters and define ranges and intervals.
This process sometimes needs other programs to assist in judgment, or relies entirely on the user's experience.
Jiang Yuan switched between multiple software programs, inputting countless numbers, before restarting the processing.
The third-generation image technician watched every step clearly—and slowly sank into despair.
"Last time, you didn't fill in this many things," Zhuang Wei asked Jiang Yuan politely.
Jiang Yuan grunted: "This video's conditions are worse. Also, the algorithm is different—some software requires more parameters."
Last time, Jiang Yuan operated on the image office's computer, where the software offered fewer algorithm choices and fewer parameters to adjust. The software installed on his own computer was naturally better suited to him.
And thus more complex.
Zhuang Wei and the other two exchanged glances, unsure what to say.
Wei Zhenguo didn't care—he sat beside them, calmly smoked two cigarettes with Wu Jun, and listened to the printer's squeaky whirring.
"I'll get it," Wei Zhenguo leapt up and rushed over to snatch the four images from the second video, moving with the speed of a seasoned thief.
This time, Jiang Yuan extracted four frames and processed them together.
This was also one reason for the high number of parameters: video processing is complex, and extracting more frames increases the chance of useful results.
After laying the several images flat on the desk and studying them for a while, Wei Zhenguo could only shake his head, take photos, and say: "I'll show these to my colleagues. Maybe they're new recruits."
Wei Zhenguo sent the photos to two WeChat groups.
Jiang Yuan packed up his gear and handed the copy drive and other items to Zhuang Wei.
The computer's hum had barely ceased when Wei Zhenguo's phone rang.
"Where did you get these images?" The voice on the other end asked immediately.
Wei Zhenguo laughed: "Can't say—the case is still under investigation. Do you recognize the person? Know their name?"
The caller hesitated for a few seconds, then said: "Looks familiar, but I'm not sure. Check the database for Zhang Wanjie. See if it's him."
"Your case?"
"Old one. Listed as wanted."
"Ha, I'll check. Thanks—I'll buy you a smoke later." Wei Zhenguo hung up.
Jiang Yuan didn't wait for him—he opened the "National Wanted Persons Information System" on the still-warm computer and typed "Zhang Wanjie."
"Armed robbery, snatch theft, burglary, highway bandit—no doubt about it," Wei Zhenguo said, grinning. His mood was like suddenly discovering a beautiful woman in a random inspection was actually a male fugitive.
"Ministry-level wanted notice," Zhuang Wei sighed. "We've caught one Ministry-level."
Ministry wanted notices are divided into two levels: Level One is the Ministry's own list of key fugitives; Level Two is compiled at provincial requests, with rewards and coordination handled by provincial bureaus.
But for grassroots police officers, catching a Level Two fugitive is already a rare achievement.
For forensic image analysis, finding fugitives is routine work. The Changyang City Forensic Image Team, for instance, always has officers searching for fugitives—just like fingerprint database searches.
Given China's current surveillance resources, anyone can become the subject of a video—only those with sharp eyes will spot them.
Of course, this is never easy, especially for those on the Ministry's wanted list—they're the ones local bureaus couldn't find, city bureaus couldn't find, then reported to provincial bureaus, then to the Ministry.
Not every step involves desperate effort, but comparisons are unavoidable.
Catching a fugitive effortlessly is like effortlessly earning a Third-Class Merit—it's inherently a bit of a flex.
Wei Zhenguo was one step away from the fugitive, yet gritted his teeth: "Only catching him counts. A lead alone is wasted."
"Should we trace through surveillance?" Zhuang Wei actively joined in.
Wei Zhenguo looked at Jiang Yuan.
"They're better at following surveillance than I am," Jiang Yuan said practically. His strength lies in image enhancement—he doesn't know how to track people through surveillance footage.
Wei Zhenguo smiled at Zhuang Wei: "Then if you can find his hideout, that'd be best."
He didn't know how Ben Guo knew Zhang Wanjie, but anyone on the wanted list is likely highly cautious—probably wouldn't even tell Ben Guo his location.
Zhuang Wei enthusiastically invited Wei Zhenguo back to his office.
In a place like Ningtai County, a wanted fugitive appears once every few years—he was eager, though perhaps also because Zhuang Wei and the others previously lacked the ability to identify fugitives; still, following surveillance footage was well within his capability.
Wei Zhenguo walked into the image office while making a call.
Ben Guo's interrogation must be intensified—he even held some hope for Ben Guo.
A thief this stupid? You really think he went to the bus station just out of friendship to pick up a fugitive? Wei Zhenguo didn't believe it.
From Ben Guo's improved living conditions, his newfound artistic flair, and the fact that theft had become a daily hobby, it's clear he's found a new way to make money—and he's still just as stupid.
Wei Zhenguo hoped to use Ben Guo to uncover more.
"Let's follow the bus station route," Zhuang Wei said, pulling up more surveillance footage from the cloud and beginning to review it frame by frame.
Wei Zhenguo joined him, occasionally helping analyze the route.
As they watched, Zhuang Wei muttered: "Jiang Yuan's luck is incredible—he keeps hitting big cases."
Wei Zhenguo, eyes fixed on the screen, smiled faintly: "Maybe his techniques are far beyond normal levels."
Zhuang Wei was stunned, then realized—he was right.
If this bus station video had been given to them, it would've been like a stone dropped in water—vanished. Tracking the fugitive would've been impossible.
Based on Zhuang Wei's understanding of forensic image analysis, unless the video was directly handed to the provincial team, the Changhe City Bureau's capability was limited. But without solid evidence, how could the video ever reach the provincial team?
Zhang Wanjie might not know the details of police systems, but his years on the run likely taught him the boundaries of safety.
Jiang Yuan precisely shattered that boundary—and easily pulled him out.
Zhuang Wei understood—and his sense of loss deepened.
Clearly surpassing colleagues and exceeding industry averages is an extremely high bar.
Realizing this, Zhuang Wei's tense muscles relaxed.
Let whoever will, will.
Tired.
End of Chapter
