In China, many families have exhausted their time and money searching for children that have gone missing for many years. Many children still haven’t been found, and their parents finally chose to give up.
There are still many parents who have lost their children and still haven’t found them. As IT professionals, what can we do to help out? How can we ensure kidnappers have no place to hide?
These problems can be solved with the continuous innovation and development of technology. Several techniques mentioned in this article may be effective against kidnappers, helping to quickly detect them and thus reducing the crime rate.
Shopping malls, transportation stations, scenic spots, cinemas and other places with large crowds and chaotic movement between all sorts of people easily become places where kidnappers start their operations. Especially during the holidays, with even greater flows of traffic, parents must be careful.
If a child goes missing, call the police as soon as possible. Let’s analyze what technical measures can help the police to quickly find potential suspects.
Nowadays, parents often buy their children a watch with a GIS feature to check their children’s whereabouts at any time.
If a kidnapper doesn’t take off the watch, this is a good way to find the child.
After learning about the location of the child, you can quickly find the nearby police through KNN query, so the kidnapper can be quickly caught.
Time is life, so query efficiency is very important. GiST index is needed here, and PostgreSQL databases provide strong support for these scenarios.
What should we do if, unfortunately, the kidnapper destroys the GIS positioning watch, or the child does not have this type of tracking device?
Humans are social animals. As the population grows, the methods of communication become more and more borderless. A huge network of relations has been formed between people-people, people-event, and people-time.
Kidnappers are also human beings, and they cannot escape this network. As the saying goes: You can’t escape the long arm of the law.
Kidnappers need to eat, shop, drink, walk, smoke, take taxis, deliver goods and use mobile phones…
Currently, many technologies can monitor these behaviors, including cameras in various corners, online cashier terminals in small shops, online terminals for taxi/online car-hailing, mobile phone location reporting, and ATM machines.
The information forms a large network. The police can conduct a radial relation derivation centered on location-related events based on the location that the child went missing, and then combine with the criminal record database in the public security system, so criminals may be found quickly.
(For example, a group of people A are circled N kilometers around the geographical location of the crime site, and then the social relation between group A and the criminal record database of the public security system is analyzed to find a breakthrough. Of course, there are more detailed conditions that can be filtered and converged to target criminals.)
After narrowing down the scope of the suspects, we should quickly start tracking them. How can we find them?
As mentioned earlier, there are a lot of cameras now, and military cameras are basically very high-definition, so the whereabouts of suspects can be captured everywhere.
For example, pedestrians photographed on cross-walks, cars in motion, crowds in shopping malls, and ATM cameras.
After extracting the portrait part, it needs to be compared with the portrait library of suspects for similarity to locate the suspect.
The suspect’s appearance often changes slightly over time, and the suspect may deliberately change their appearance. How can we match a suspect across many image libraries?
Generally, the suspect may have more than one picture in the image library. While there may be many pictures, I suppose the face should not change that much.
If the child’s image can be captured, the child’s image recognition rate is also very high.
In short, kidnappers have nowhere to hide.
In addition, the consumption behavior of the suspect, the location reported by mobile phones and so on, are all helpful information for catching the suspect.
You might say, the suspect does not have to spend money, use a phone, or withdraw money personally, in which case they’d need someone to do it for them, which goes back to the relation derivation analysis. You know, they wont escape the long arm of the law.
If we have already taken a picture of the suspect in a car from a certain camera, but the car is deliberately disguised (for example, the license plate number is partially blocked), how can we find the car quickly?
Fuzzy search is involved here.
We talked a lot about how to find suspects through data analysis, and data processing. However, we have to know that we are facing a huge amount of data. How many cameras, mobile phones, ATM machines, consumption terminals and taxis are there in the country? There are a huge number of devices producing an even more massive amount of data.
Therefore, due to the huge amount of information generated in society, how can we quickly track the whereabouts of suspects using the vast amount of information?
Stream processing technology helps solve this challenge. For example, we have locked the scope of the suspect. From a technical point of view, it needs to filter out the suspect’s information from the massive amount of information generated in real time, or to trigger a notification immediately to inform the public security officers when the information matching the suspect is hit.
Those who have played CS, especially those who are good snipers, must be familiar with the snap shot. This basically means to shoot a moving target.
The same is true for suspects. For example, we have used technical means to locate the whereabouts of the suspect.
The escape route of the suspect can be depicted, but when you arrive at the scene, the suspect may be at the next location.
If we can deduce where the suspect might go next, then we can ambush the suspect in advance similar to landing a snap shot in the game.
Two technologies are involved here. One is data mining (some regression analysis technologies may be used), and the other is path planning.
Public security officers can also artificially create some inducements, such as setting obstacles, and misleading traffic direction.
Again, you can’t escape the long arm of the law. The suspects will eventually be arrested, and the fight against kidnappers will be further strengthened with the progress of technology. Eliminating kidnapping from the world is a valiant goal of IT workers.
The seller and the buyer commit the same crime.
After the abducted children are found, how can we find their relatives?
Image matching is a method. For example, the Haar Wavelet algorithm mentioned earlier is used to compare pictures of the children with pictures of the missing children in the public security database, and find the registration information of the missing children, to find their relatives.
However, some children may have been missing for many years, and their appearance has certainly changed a lot over the past few years, so they cannot be completely identified by pictures alone.
Fortunately, blood samples can be used to identify ancestry, and now we compare the DNA similarity to identify relatives. After finding a lost child, we extract the DNA of the child and compare it with the DNA bank for parents who reported that their children were abducted (there may be many of these families, estimated to be in the millions).
Find the family with the highest similarity, make further confirmation, and finally help the child reunite with relatives.
DNA matching is also achieved by technical means.
At this point, the war between IT workers and kidnappers has come to an end, but we still have a lot of advanced technologies not introduced here. We will never go easy on kidnappers.
PostgreSQL is a database “full of love”. Learn it and reward society with your heart.