純粹翻譯分享，原文來自photometrics如下
https://www.photometrics.com/learn/camera-basics/how-is-an-image-made

How Is An Image Made? --- 圖像是如何制作的？

Introduction

Cameras are incredible tools that allow us to capture and make sense of the visible world around us. Most mobile phones made today come with a camera, meaning that more people than ever are becoming familiar with camera software and taking images. But one of the largest applications of cameras is for scientific imaging, in order to take images for scientific research. For these applications, we need carefully manufactured scientific cameras.

What Is Light?

The most important aspect of a scientific camera is the ability to be quantitative, measuring specific quantities of something. In this case, the camera is measuring light, and the most basic measurable unit of light is the photon.

科學(xué)相機最重要的方面是定量的能力，測量特定數(shù)量的某物。在這種情況下，相機正在測量光，光的最基本的可測量單位是光子。

Photons are particles that make up all types of electromagnetic radiation, including visible light and radio waves, as seen in Figure 1. One of the most important part of this spectrum for imaging is visible light, which ranges from 380-750 nanometers, as seen in the insert of Figure 1.

光子是構(gòu)成所有類型電磁輻射的粒子，包括可見光和無線電波，如圖 1 所示。成像光譜中最重要的部分之一是可見光，其范圍為 380-750 納米，如圖 1 的插頁所示。

What-is-light.png

Figure 1: The electromagnetic spectrum. This spectrum indicates what form of radiation is produced from photons at different wavelengths and frequencies, with photons of higher frequency having higher energy and lower wavelength, and vice versa. With increasing wavelength/decreasing frequency/decreasing energy, the spectrum includes gamma rays (Greek letter for gamma: γ), x-rays, ultraviolet (UV), visible light (more detailed spectrum shown in insert), infrared (IR), microwave, standard radio waves (including frequency modulation FM and amplitude modulation AM commercial radio frequencies) and long radio waves. Wavelength is shown in magnitudes of 10 in meters, frequency in magnitudes of 10 in Hz. For the visible spectrum, different wavelengths produce different colors, including violet (V, 380-450), blue (B, 450 495), green (G, 495-570), yellow (Y, 570 590), orange (O, 590-620), red (R, 620-750), all wavelengths in nanometers (nm). Image from Wikimedia Commons.

圖 1：電磁頻譜。該光譜表示不同波長和頻率的光子產(chǎn)生哪種形式的輻射，頻率較高的光子具有較高的能量和較低的波長，反之亦然。隨著波長的增加/頻率的降低/能量的降低，光譜包括伽馬射線（伽馬的希臘字母：γ）、X 射線、紫外線 （UV）、可見光（插頁中顯示的更詳細(xì)光譜）、紅外線 （IR）、微波、標(biāo)準(zhǔn)無線電波（包括調(diào)頻 FM 和調(diào)幅 AM 商用無線電頻率）和長無線電波。波長以 10 的幅度表示，以米為單位，頻率以 10 的幅度表示，以赫茲為單位。對于可見光譜，不同的波長會產(chǎn)生不同的顏色，包括紫色（V，380-450），藍色（B，450 495），綠色（G，495-570），黃色（Y，570 590），橙色（O，590-620），紅色（R，620-750），所有波長均以納米（nm）為單位。圖片來自Wikimedia Commons。

As microscopes typically use visible, infrared (IR) or ultraviolet (UV) light in the form of a lamp or laser, a scientific camera is essentially a device that needs to detect and count photons using a sensor.

由于顯微鏡通常以燈或激光的形式使用可見光、紅外線 （IR） 或紫外線 （UV），因此科學(xué)相機本質(zhì)上是一種需要使用傳感器檢測和計數(shù)光子的設(shè)備。

Sensors

A sensor for a scientific camera needs to be able to detect and count photons, and then convert them into electrical signals. This involves multiple steps, the first of which involves detecting photons. Scientific cameras use photodetectors, where photons that hit the photodetector are converted into an equivalent amount of electrons. These photodetectors are typically made of a very thin layer of silicon. When photons from a light source hit this layer, they are converted into electrons. A layout of such a sensor can be seen in Figure 2.

科學(xué)相機的傳感器需要能夠檢測和計數(shù)光子，然后將其轉(zhuǎn)換為電信號。這涉及多個步驟，其中第一個步驟涉及檢測光子。科學(xué)相機使用光電探測器，其中撞擊光電探測器的光子被轉(zhuǎn)換為等量的電子。這些光電探測器通常由非常薄的硅層制成。當(dāng)來自光源的光子撞擊該層時轉(zhuǎn)化為電子。這種傳感器的布局如圖2所示。

Sensor.png

Figure 2: A cross-section of a camera sensor. Light first hits the microlens (top of image), which focuses the light onto the silicon pixel (at the bottom of the image). The sensor area outside of this light path is full of integrated electronics and wiring.

圖 2：相機傳感器的橫截面。光線首先照射到微透鏡（圖像頂部），微透鏡將光線聚焦到硅像素（在圖像底部）。該光路之外的傳感器區(qū)域充滿了集成的電子設(shè)備和布線。

Sensor Pixels

However, having just one block of silicon would mean detection was possible, but not localization. By separating the silicon layer into a grid of many tiny squares, photons can be both detected and localized. These tiny squares are referred to as pixels, and technology has developed to the point where you can fit millions of them onto a sensor. When a camera advertises as having 1 megapixel, this means the sensor is an array of one million pixels, a 1000×1000 grid.

然而，只有一塊硅意味著可以進行檢測，但不能進行定位。通過將硅層分離成許多小方塊的網(wǎng)格，可以檢測和定位光子。這些微小的方塊被稱為像素，技術(shù)已經(jīng)發(fā)展到可以將數(shù)百萬個正方形安裝到傳感器上的地步。當(dāng)相機宣傳為具有 100 萬像素時，這意味著傳感器是一個 100 萬像素的陣列，即 1000×1000 網(wǎng)格。

In order to fit more pixels onto sensors, pixels have become very small, but as there are millions of pixels the sensors are still quite large in comparison. The Prime BSI camera has 6.5 μm square pixels (42.25 μm2 area) arranged in an array of 2048 x 2048 pixels (4.2 million pixels), resulting in a sensor size of 13.3 x 13.3 mm and a diagonal of 18.8 mm. Meanwhile, the Prime 95B has the same 18.8 mm diagonal sensor, but with 11 μm square pixels (area of 121 μm2) in a 1200 x 1200 array (1.4 million pixels). The Prime 95B, therefore, has fewer sensor pixels (decreasing the maximum imaging resolution), but each pixel is 3x larger in the area (increasing the sensitivity).

為了在傳感器上安裝更多的像素，像素變得非常小，但由于有數(shù)百萬像素【像素就是像元pictureelement】，相比之下，傳感器仍然相當(dāng)大。 PrimeBSI相機具有6.5μm方形像素(42.25μm2區(qū)域)，排列成2048x2048像素(420萬像素)的陣列，傳感器尺寸為13.3x13.3mm，對角線為18.8mm。 Prime95B相機具有11μm方形像素(面積為121μm2)，排列成1200x1200陣列(140萬像素)的陣列，傳感器尺寸為13.3x13.3mm，對角線為18.8mm。 后肢傳感器像素較少(降低了最大成像分辨率)，但每個像素的面積增加了3倍(提高了靈敏度)。

Making sensor pixels smaller allows for more to fit on a sensor, but if pixels become too small they won’t be able to detect as many photons, which introduces the concept of compromise in camera design between resolution and sensitivity. One option to consider is binning, which is discussed in a separate article. Due to these reasons the overall sensor size, pixel size, and the number of pixels are carefully optimized in camera design. When deciding which scientific camera to get, pixel size is a vital metric that is important to consider.

傳感器像素變小可容納更多像素，但如果像素變得太小將無法檢測到盡可能多的光子，這在相機設(shè)計中引入了分辨率和靈敏度之間折衷的概念。可以考慮的一個選項是binning分箱，這將在另一篇文章中討論。由于這些原因，在相機設(shè)計中，傳感器的整體尺寸、像素尺寸和像素數(shù)量都經(jīng)過了精心優(yōu)化。在決定購買哪種科學(xué)相機時，像素大小是一個重要的指標(biāo)，需要考慮。

Generating An Image

When exposed to light, each pixel of the sensor detects how many photons come into contact with it. This gives a map of values, where each pixel has detected a certain number of photons. This array of measurements is known as a bitmap and is the basis of all scientific images taken with cameras, dependant on the signal level of the experiment and application. The bitmap is accompanied by the metadata, which contains all the other information about the image, such as the time it was taken, camera settings, imaging software settings, and microscope hardware information.

當(dāng)暴露在光線下時，傳感器的每個像素都會檢測到有多少光子與之接觸。這給出了一個值圖，其中每個像素都檢測到一定數(shù)量的光子。這種測量數(shù)組被稱為位圖，是相機拍攝的所有科學(xué)圖像的基礎(chǔ)，取決于實驗和應(yīng)用的信號電平。位圖附帶元數(shù)據(jù)，其中包含有關(guān)圖像的所有其他信息，例如拍攝時間、相機設(shè)置、成像軟件設(shè)置和顯微鏡硬件信息。

The following are the processes involved in generating an image from light using a scientific camera:

1. Photons hit the sensor are converted into electrons (called photoelectrons). 撞擊傳感器的光子被轉(zhuǎn)化為電子（稱為光電子）。
   • The rate of this conversion is known as quantum efficiency (QE). With a QE of 50%, only half of the photons will be converted to electrons and information will be lost. 這種轉(zhuǎn)換的速率稱為量子效率 （QE）。在50%的QE下，只有一半的光子會轉(zhuǎn)化為電子，信息會丟失。
2. The generated electrons are stored in a well in each pixel, giving a quantitative count of electrons per pixel 生成的電子存儲在每個像素的孔well中，從而給出每個像素的電子定量計數(shù)
   • The maximum number of electrons that can be stored in the well is known as the full well capacity, which determines the dynamic range of the sensor. 阱中可以存儲的最大電子數(shù)稱為全阱容量，它決定了傳感器的動態(tài)范圍。
3. The electron charge in each pixel’s well is amplified into a readable voltage, this is the analogue signal. 每個像素阱中的電子電荷被放大成可讀的電壓，這就是模擬信號。
4. The analogue signal is converted from a voltage into a digital signal with an analogue to digital converter (ADC). This arbitrary digital signal is known as a grey level, as most scientific cameras are monochrome. 模擬信號通過模數(shù)轉(zhuǎn)換器（ADC）從電壓轉(zhuǎn)換為數(shù)字信號。這種任意數(shù)字信號被稱為灰度電平，因為大多數(shù)科學(xué)相機都是單色的。
   • The rate of this conversion is known as gain. With a gain of 1.5, 100 electrons are converted to 150 grey levels. 這種轉(zhuǎn)換的速率稱為增益gain。增益為 1.5 時，100 個電子被轉(zhuǎn)換為 150 個灰度級別。[gain解釋不夠]
5. The bit-depth of the camera determines how many grey levels are available for the signal to be converted into, a 12-bit camera has 4096(2^12) available grey levels, a 16-bit camera has (2^16). 相機的位深度決定了有多少灰度級別可用于轉(zhuǎn)換信號，12位相機有4096(2^{12)個可用灰度級別，16位相機有65,536(2}16)。
   • The bit depth also determines the full well capacity and therefore the dynamic range. 位深度還決定了整井容量，從而決定了動態(tài)范圍。
6. The map of grey levels is displayed on the computer monitor in the imaging software as an image. 灰度圖以圖像形式顯示在成像軟件的計算機顯示器上。
   • The generated image depends on the software settings, such as brightness, contrast, etc. 生成的圖像取決于軟件設(shè)置，例如亮度、對比度等。

These steps are visualized in Figure 4.

Generating-an-image.png

Figure 4: The process of taking an image with a scientific camera. Photons impact the sensor, which produces photoelectrons, the rate of production is known as quantum efficiency. These electrons go into the well of each pixel and are counted, amplifed and converted into grey levels by an analogue to digital converter (ADC). These grey levels are then displayed on a computer monitor, with the image appearance controlled by display settings in the software (contrast, brightness, etc).

圖4：使用科學(xué)相機拍攝圖像的過程。光子撞擊傳感器，產(chǎn)生光電子，產(chǎn)生速率稱為量子效率。這些電子進入每個像素的阱，并由模數(shù)轉(zhuǎn)換器(ADC)進行計數(shù)、放大并轉(zhuǎn)換為灰度電平。然后，這些灰度級別顯示在計算機顯示器上，圖像外觀由軟件中的顯示設(shè)置(對比度、亮度等)控制。

In this manner, photons are converted to electrons, which are converted into a digital signal and displayed as an image. These main stages of imaging with a scientific camera are consistent across all modern camera technologies, but there are several different types of sensor architecture and design. 以這種方式，光子被轉(zhuǎn)換為電子，電子被轉(zhuǎn)換為數(shù)字信號并顯示為圖像?？茖W(xué)相機成像的這些主要階段在所有現(xiàn)代相機技術(shù)中都是一致的，但有幾種不同類型的傳感器架構(gòu)和設(shè)計。

Types Of Camera Sensor

Camera sensors are at the heart of the camera and have been subject to numerous different iterations over the years. Researchers are constantly on the lookout for better sensors which can improve their images, bringing better resolution, sensitivity, field of view, and speed. The three main camera sensor technologies are:

相機傳感器是相機的核心，多年來經(jīng)歷了許多不同的迭代。研究人員一直在尋找更好的傳感器，以改善他們的圖像，帶來更好的分辨率、靈敏度、視野和速度。三種主要的相機傳感器技術(shù)是：

• Charge-coupled device (CCD)電荷耦合器件 （CCD）
• Electron-multiplied charge-coupled device (EMCCD)電子倍增電荷耦合器件 （EMCCD）
• Complementary metal-oxide-semiconductor (CMOS)互補金屬氧化物半導(dǎo)體 （CMOS）

Each of these sensors is discussed in detail in our next article, Camera Sensor Types