I. Principles of camera imaging

Industrial cameras are a key component in machine vision systems, and their most essential function is to convert optical signals into ordered electrical signals.Choosing the right camera is also an important part of machine vision system design. The choice of camera not only directly determines the collected image resolution, image quality, etc., but also directly related to the operating mode of the entire system.The imaging principles of industrial cameras can be divided into two types: ‌CCD and ‌CMOS. Each type has its own unique working methods and advantages

2. Industrial camera parameters

1. Definition of industrial camera parameters

1.1 shutter-shutter

Shutter is a way to control exposure time. Common electronic shutters are global exposure and roller shutters.

Global shutter: When exposed, all pixels on the sensor turn on exposure at the same time and end at the same time, image the object's state at a certain moment, suitable for shooting moving objects.

Roller shutter: Exposure is to open exposure line by line. There is a delay in opening exposure between different lines, so it is not suitable for shooting moving objects.

Note: The exposure time should not be too long, as it is easy to cause blur.For moving objects, the shorter the exposure time, the clearer the image of the object, but it is easy to have darker the image because the exposure time is short, the amount of light entering is small, and the gain of the photosensitive chip is small.

1.2 Exposure-exposure refers to the process in which the image sensor performs photosensitive during the operation of an industrial camera, the CCD/CMOS collects photons and converts them into charge, and the charge amount is constantly accumulated.

Exposure time refers to the time interval between the camera from the shutter opening to the closing. The exposure time is short and the charge accumulates less; after the exposure is over, CCD/CMOS removes the charge in a certain way, and the exposure from light has a great impact on the quality of the photo. If the exposure is over, the photo will be too bright and the image details will be lost; if the exposure is insufficient, the photo will be too dark and the image details will be lost.Therefore, under the premise of not being exposed, the exposure time should be increased, the signal-to-noise ratio of the image should be improved, and the image should be clearer.

Factors that affect exposure:

Aperture: The aperture controls the size of the path through which light enters.The larger the aperture, the greater the luminous flux entered; otherwise.The aperture size is shown below.

Exposure time: The time when the photosensitive element uses to expose light when light enters, that is, the shutter.

Gain: Generally, when the signal is weak, it is used without wanting to change the aperture and exposure time.Generally, the camera gain will generate a lot of noise, so try to use it as little as possible.

So, exposure time and aperture are generally used in combination to achieve the required exposure.For example, for high-speed moving objects, the exposure time is reduced and the aperture is increased so that the object details are more and not too dark.

1.3 Gain-gain

***Gain is usually used when the signal is weak but does not want to increase the exposure time. Generally, the camera gain produces a lot of noise.***With short exposure time, less charge accumulation on CCD/CMOS, the gain needs to be strengthened; conversely, the gain must be weakened.

1.4 Photosensitive chip type (CCD/CMOS)

CCD (charge coupled device) chip and CMOS (complementary metal oxide semiconductor) chip both convert optical signals (photons) into electronic signals (electrons).

Image source: CCD/CMOS Picture source: Infringement delete

CCD camera is more suitable for high-quality or moving objects images, and the CCD imaging quality is higher than that of CMOS; such as the above CCD camera.

CMOS cameras are more suitable for image acquisition with static or high acquisition speed requirements. The CMOS cost is lower than that of CCD and has the advantage of acquisition speed.

The appearance of the two pictures is not different, mainly because of the difference in the internal photosensitive chips. This is only used for schematic purposes.

1.5 Size of photosensitive chip (target surface)

Size of photosensitive chip (target surface) is literally meaning, which refers to the length and width of the camera photosensitive chip. Usually camera manufacturers use inches such as units (such as 2/3’’), and actually need to be converted into units of mm. We usually use the following formula:

1.6 Resolution

The camera's sensor is made up of many pixels arranged in the form of a matrix, and the resolution is represented by pixels in the horizontal and vertical directions.For example, if the resolution is 1920*1080, then the number of physical cells in the width direction is 1920, and the number of cells in the height direction is 1080.The higher the resolution, the higher the pixel count of the image after imaging and the clearer the image.The resolution of the camera does not necessarily have to be greater, and you need to choose from the cost and project needs.

Common surface array camera resolutions include:

640 *480 — 300,000 pixels; 1280 *960 — 1.3 million pixels; 1440 *1080 — 1.6 million pixels;

1600 *1200 — 2 million pixels; 1920 *1080 — 3 million pixels; 2590 *1940 — 5 million pixels, etc.

Common linear array cameras (generally refer to the number of pixels in the horizontal direction, because the number of pixels in the vertical direction is generally 1) have the following resolutions:

1024-1K      2048-2K 6144-6K

1.7 Cell size

Cell size refers to the length and width of the pixel.When the photosensitive chip size is certain, the smaller the cell size, the larger the resolution, which is more conducive to the feature expression and defect detection of the image.1.8 frame rate

Frame rate refers to the number of frames collected per second in FPS or Hz.The frame rate determines the number of pictures collected per unit time, that is, the speed of pictures collected.

Plane array Sensor: fps frame rate, maximum number of pictures collected per second;

Line array Sensor: Example 16kHz line frequency, maximum number of rows collected per second, maximum image resolution is resolution (width)* line frequency (Example 2048x16k).

1.9 Cell Depth

The depth of a cell (pixel) refers to the number of bits of storage space occupied by storing the grayscale value of each pixel. The most common cell bit depths are 8 bits, 16 bits, 24 bits, and 32 bits.The depth of the cell/pixel determines the number of colors each cell can represent in the color image, or determines that the grayscale image may result in the number of grayscale levels.

Resolution and cell depth determine the memory size of the image.

1.10 Commonly used camera interface

GIGE Gigabit network interface:

GIGE Gigabit network protocol is stable and easy to use. It can work when connected to a gigabit network card

The transmission distance is long and can be transmitted 100 meters.

It can be used at the same time, and the CPU usage is small.

USB interface:

The early USB2.0 interface was convenient to connect, and almost all computers were equipped with a USB interface without a collection card.

The USB2.0 interface has a slow transmission rate, and the transmission process requires the CPU to participate in management, which occupies and consumes a lot of resources.

The USB2.0 interface generally does not have fixing screws, the interface is unstable, and there is a risk of loosening on sports equipment.

USB3.0 has added two new data cables on the basis of 2.0, which are backward compatible, solving the problem of slow transmission speed, but the transmission distance is still close and cannot exceed 5m.

Camera Link interface:

You need to purchase an additional image acquisition card (basically around 10,000 yuan), which is very costly..

Camera Link interface cameras are relatively rare in actual applications.

Using the LVDS interface standard, it has a fast speed, strong anti-interference ability and low power consumption.The transmission distance is close.

1.11 Camera triggering method

Software triggering mode: You can choose (software control) when dynamic detection and product triggering signals through continuous motion.

Hardware trigger mode: Select (sensor or mechanical switch control) when high-speed dynamic detection and product trigger signals through high-speed motion.

Continuous acquisition mode: Select when static detection and continuous product movement cannot trigger signals.

1.12 Camera and lens interface types

General cameras are C/CS interfaces, and attention should be paid to the corresponding lens.If there are lenses with other interfaces, you should also consider the camera interface.Industrial cameras with C-type interfaces cannot use CS-port lenses. CS-port cameras can add an adapter to use C-port lenses.

F interface lens is the interface standard of Nikon lens, so it is also called Nikon port, which is also a commonly used type in industrial industrial cameras.

V interface lenses are mainly used by Schneider lenses. They are generally used for lenses with larger target surfaces or special purposes in industrial cameras.

Pay attention to the interface of the camera's lens, otherwise the purchased camera may be incompatible with the existing lens.

1.13 Camera color selection

If the project is related to the image color, use a color camera, otherwise it is recommended to use black and white, which is not only low in cost but also high in accuracy.

1.14 Drag

When shooting moving images, the same object is repeatedly imaged on the picture.The reason is that the image formed by moving objects on the photosensitive chip moves more than one cell size during exposure time.—Based on this line, we can calculate the required exposure time to avoid distortion.

2. Camera selection

The first thing to choose a camera is to clarify the requirements: 1. Detection accuracy 2. Field of view (FOV) 3. Is the object/camera moving?

Example: It is known that the customer's lens size is 1/3, the interface is a CS interface, the field of view is 12x10mm, and the required accuracy is 0.02pixel/mm. What kind of camera should we choose?

2.1 Measurement accuracy and field of view->Determine the resolution of the camera

Under normal circumstances, the pixel accuracy of the camera must be higher than the accuracy required by the system. Therefore, in actual industrial applications, the actual resolution is 3-5 times the theoretical resolution, and generally 4 is taken.

Theoretical resolution (lowest resolution from)=(long field of view/theoretical accuracy)*(short field of view/theoretical accuracy)=(12/0.02)x(10/0.02)=600x500

Actual resolution =600x500x4=1200x1000 (and then select it nearest)

2.2 Object motion method->Shutter method of the camera

Select a global shutter for moving objects; select a rolling shutter for stationary objects.

2.3 The motion speed of an object->The frame rate of the camera

Generally speaking, the lower the resolution of the camera, the higher the frame rate for the same interface, and the higher the resolution, the lower the frame rate.Frame rate * resolution ≤ bus bandwidth, that is, when the interface is constant (the bus bandwidth has been determined) and the resolution is constant, the frame rate also has its maximum value.That is, if you want the camera to have a fast resolution and a high frame rate, you need to find a bus with a larger bandwidth, that is, the camera's output interface.

According to the detection speed, the frame rate of the camera must be greater than the speed of the object's movement, and must be completed within the exposure and transmission time of the camera.

2.4 Detection content->Color/Black and White Camera

If color recognition, color defect detection and other processing are related to color, then choose a color camera, and the others use black and white cameras.The same resolution, black and white cameras have high accuracy and good contrast, etc., and are cheap.

For stationary detection or generally low-speed detection, surface array cameras are preferred, and line array cameras are considered for special applications such as large-format high-speed movement or rollers.

2.5 Lens and Working Scenarios->Sensor target surface size of the camera

The size of the target surface directly affects the size of the shooting field of view.At the same working distance and the same lens, the sensor size is large and can capture a larger field of view.(Just imagine the small hole imaging model).

2.6 Frame rate, transmission distance, economy->Camera interface

The common camera interfaces on the market are USB2.0, USB3.0, and Ganzone network interfaces.

USB2.0 theoretical bandwidth is 450Mb/s, transmission distance is 5m.

USB3.0 Theoretical bandwidth is 5Gb/s, transmission distance is 5m.

The theoretical bandwidth of the Gigabit network interface is 1Gb/s, and the transmission distance is 100m.

Camera Link interface has a fast transmission speed and a close transmission distance. It requires a supporting image acquisition card, which is costly.