Chapter 58

[ha ha, I always feel familiar with the internal structure of the small box made by ridorf you sent me.

The structure of this thing is similar to the principle of the old automatic tracking missile. I know what ridov thinks.

To get back to business, the human eye's ability to distinguish brightness is very weak.]

Professor Frank was worried that Liang Shicheng didn't understand his ideas and continued to explain

[as other color blind animals, when they cannot distinguish objects by color, that is, black-and-white vision.

But their eyes are better than ours sometimes. Animals distinguish objects by brightness difference.

This is especially reflected at night. People's night vision ability is very weak.

In fact, there is not no light at night, but some night light will be leaked, although it is dark.

But many animals can see very black, extremely black, slightly black, relatively white black and so on.

The degree of darkness in different places is not exactly the same. They can be clearly distinguished, so they have the so-called night vision ability.

However, the degree of black that the human eye can see is limited, so it is difficult to distinguish between slightly black and very black.

Brightness or darkness is actually the number of photons passed per unit area. Theoretically, it can be subdivided infinitely from white to black.]

Although Liang Shicheng doesn't know much about biology, he still knows something about the defect of human eye resolution brightness because of his exposure to photography, and keeps up with it.

Well, those high dynamic range cameras, which are more capable of subdividing the brightness level, do not need to increase exposure. After shooting the night, adjust the gamma curve to see a clear picture.

Professor Frank didn't understand what Liang Shicheng was saying, so he shook his head and smiled. For a professor who makes a living cleaning the toilet, many concepts are clear, but he forgot the professional words.

Then Professor Frank went on to add that, as the party who took the initiative to answer the question, he obviously paid more attention to being easy to understand.

[our insensitivity to brightness leads to extremely poor night vision. In fact, the impact also exists during the day,

Just because the rich colors during the day make up for this short board.

Answer me a question on a sunny day.

What is the relationship between the brightness of the sun, sky, clouds, and objects exposed to the sun on the ground, such as large trucks.]

Liang Shicheng immediately answered, which seemed to be a simple question.

[they all look the same and bright with their eyes. I see what you mean. It's not true, is it.]

Professor Frank explained in detail,

[yes, you should remember that from a physical point of view, the light source must be much brighter than other objects.

Let me give an example.

It should be night over there. Look at the glass. Is the picture of the room reflected by the window glass very dark, but the light in the glass is very bright, almost like the real light in the room?]

Liang Shicheng did as he was told, but in fact he had noticed this at ordinary times, especially when playing with mobile phones in the car,

That's true

Professor Frank soon picked up the previous topic,

[in fact, you see the reflected room in the window, which is more real in some cases.

The reflectivity of transparent glass windows is very low, which can be weakened by dozens of times, which is equivalent to a mirror with reduced brightness. Therefore, most objects in the room are relatively dark in the windows.

But only the light is different. The real brightness of the light is very high, weakened dozens of times, and it is still very bright in our eyes.

That's the problem. Pay close attention to the lights in the glass and the lights in the room. Is there a big gap between them? There seems to be no gap of more than ten times.

That is because the image seen by the human eye is after PS, which will darken the too bright places and slightly brighten the dark places,

Let us not feel the real brightness gap.

The lights and objects in the room you see directly are almost bright, but it's fake.

The real situation is closer to what you can see in the glass. The light source is very bright and the illuminated object is dark.

As for the answer to the previous question, the sun is brighter than the sky, the sky is much brighter than clouds, and the sky is much brighter than big trucks.

Then you should remember that the brightness of the light source is always much brighter than that of the object that does not emit light and is passively illuminated.]

[I remember one thing. It's about rattlesnake, unlocking mobile phone face recognition.

Generally speaking, many mobile phones began to open a function of using face unlocking instead of password, which originally had high requirements for security.

But hackers soon found a way to crack it. That is, they can muddle through by getting a picture of the mobile phone owner. For mobile phones with only one front camera, this move is difficult to solve.

But if it is not solved, it means security risks. I paid attention to ridorf's account and found that he provided a relaxed and perfect solution to mobile phone manufacturers.

That is to add a lens that imitates the eyes of a rattlesnake. Although the rattlesnake has poor eyesight, it can see the light emitted by creatures clearly.

Any animal will emit light of a specific frequency, but it can't be seen by human eyes.

The cost of this method is not high. If someone puts a photo in front of the lens, because the photo is not a creature and will not emit specific light on people, it will soon be able to tell that it is not a real person.

As soon as you said that ridorf was tossing the mouse again, I remembered that although the mouse's eyes could not see a clear picture, they could recognize the brightness more accurately than human beings.

In the eyes of mice, the sky is very bright and people are black. For them, face recognition is not necessary. When they see a large lump of black blocking the light, they can naturally detect the danger.

I think the idea of the small box is like this. The small hole is a camera imitating mouse eyes. Although it can't recognize pictures, it can accurately identify brightness.

The sky is a very bright light source, and the big truck is a lump of black in the rat. No matter it is painted with imitation sky camouflage, it can only deceive people. At the same time, it can also deceive the general camera with parameters close to human eyes.

But I can't fool the mouse's eyes.]

Hearing this, Liang Shicheng was stunned. Although he had roughly figured out the principle and guessed the method.

However, it is only with such an explanation that I feel that I have been engaged in image recognition for so many years and know the difficulty.

It is very difficult to teach computers which is the sky, which is the ground and which is the person in a colorful world.

In particular, I'm afraid that when people wear clothes painted by the sky, AI will confuse the clothes with the sky.

This is why many companies insist on using lidar, but lidar is too expensive.

But I never thought of making changes in the hardware. As long as I put on another pair of glasses, the sky is bright, the people are dark, and the big truck is also dark. It is very obvious and easy to identify.

The idea is frighteningly simple. You only need to change an "eye" to calculate brightness, ignore color and other information, and you can do it without high resolution.

Inspired by Liang Shicheng, he finally solved all the problems after adding a special lens to identify the brightness.

At the same time, Liang Shicheng lamented that Mr. ridorf, who claimed that he was not proficient in software technology, was really a rare generalist.