Machine Vision-Camera

2. In the case of high-speed moving objects, the image should be collected by a progressive scan camera or a field acquisition method using an interlaced camera.
4. The line array camera is suitable for the large size of the target object, or small defects, such as: Printing quality inspection, PCB testing, cloth or cotton detection, grain (grain, fruit, etc.) detection.
6. Main performance comparison of CCD and CMOS
   • Full-trap capacity differences: Since each pixel of the CMOS sensor includes a photosensitive diode, amplifier and readout circuit, and the entire sensor also includes an addressing circuit and a/d, so that the photosensitive region of each pixel is much smaller than the surface area of the pixel itself, so that in the case of the same pixel size, The full-trap capability of CMOS sensors is lower than that of CCD sensors.
   • Cost variance: Because CMOS sensors use the most common CMOS process in general semiconductor circuits, peripheral circuits such as AGC, CDS, timing, or DSP can be easily integrated into the sensor chip, thereby saving the cost of peripheral chips.
   • CMOS sensors can be randomly addressed, which makes it very easy to read only parts of the queue that are interested in them, increasing the frame rate.
   • Noise difference: Because each sensor diode of the CMOS sensors is paired with an amplifier, and the amplifier is analog, it is difficult to keep the results of each amplifier consistent, so the CMOS sensor has a lot more noise than a CCD sensor with only one amplifier at the horizontal register output. affect image quality.
   • Power difference: The CMOS sensor consumes less power than the CCD sensor.
8. Maximum line frequency
   • Indicates the maximum number of lines that a line array camera can output per second.
10. Maximum frame rate
    • The maximum number of frames a polygon array camera can capture and output per second. Factors that limit the maximum frame rate are exposure time, data readout time (the time the image data is transmitted from the sensor to the camera's internal FIFO or memory), data transfer time (the time the image data is from the camera's internal FIFO or memory to the PC), the pixel depth, and the signal output format.
      The manual usually lists the frame rate at the maximum resolution. If partial scanning or binning technology is used, the output frame rate of the camera is increased. Of course, the former Narrows the acquisition field, which reduces the resolution.
12. Pixel depth
    • Pixel depth refers to how many bits per pixel are represented. In general, the number of bits per pixel and the ability to express image details is strong.
14. Signal-to-noise ratio
    • The Snr is the ratio of the measured signal to the total noise measured.
16. Dynamic Range
    • The dynamic range is used to describe the level of gray that each pixel can distinguish. It is the ratio of the noise of the camera output to the saturation voltage (maximum output level). The wide dynamic range allows the details of the very bright and very dim parts of the scene to be clearly displayed at the same time, and the details of the background and the details of the face can be seen very clearly when the person is standing back in very bright light.
18. Speed
    • The camera's sensitivity (responsiveness) defines the sensitivity of the camera to the fixed exposure, which is related to the wavelength of the light wave being irradiated.
20. Image Acquisition Method
    • Image Acquisition method is mainly refers to (1) when to start collecting? (2) The length of exposure? (3) How to end the collection?
    • There are usually two kinds of acquisition and triggering of external hardware signals. Software acquisition is a computer to the camera to start the acquisition of commands, can be single-frame acquisition or multi-frame continuous acquisition, external trigger acquisition is triggered by the outside trigger signal acquisition, the signal can be rising or falling edge trigger or level trigger.
    • Exposure time, the software command acquisition is through the register programming to achieve different lengths of exposure, the external triggering of the acquisition is also through the Register programming set implementation, level trigger is triggered by the level of the duration of control.
    • The acquisition end control is mainly for the software acquisition, the acquisition command is automatically cleared after the end of the single frame acquisition, and the collection command to be issued by the computer to clear the command is collected continuously.
22. Exposure (shutter)
    • The shutter is the control of the exposure time. The traditional camera is the exposure control through the mechanical shutter to the light which shines on the film. In CCD or CMOS cameras, exposure control is achieved using an electronic shutter. The light that shines on the sensor is continuously photoelectric converted, the principle of the electronic shutter is only to store a certain exposure time signal charge, and then output.
    • The common electronic shutter has two modes of operation: Global shutter (frame exposure), rolling shutter (progressive exposure).
    • The Global shutter is the simultaneous exposure of all pixels in the sensor array, and the exposure period is determined by the pre-set shutter time. Some CMOS sensors use the rolling shutter, and all pixels on the sensor are not exposed simultaneously, but the pixels on the same line are exposed simultaneously, and the exposure start time is different. The first line of the image is first exposed, first ending, the subsequent line start and end exposures are slightly delayed than the previous line, overlapping each other, the exposure time of each row is the same, and the delay between the lines is constant.
    • as the rolling shutter mode, the whole image exposure needs a certain time, for high-speed moving objects, will cause blurred image, deformation .
    • Comparison of effects
24. Binning
    • The binning is a photosensitive pixel that is merged in the sensor and then output. The merging of pixels can be between rows, or between columns, so there are vertical binning and horizontal binning. The vertical binning is to move the adjacent two lines into the horizontal shift register, and then shift the output together, and the horizontal binning is to move the neighboring pixels in the horizontal shift register to the amplifier, and no longer between adjacent pixels.
      It is obvious that the neighboring pixel binning, the signal charge increases one times, the camera sensitivity to light increases by 6dB, the SNR increases by 3dB. Of course, the image spatial resolution after binning is reduced.
26. Blooming
    • Blooming refers to the scene in which there is a very bright part, in the photoelectric conversion process, the sensor corresponding to the very bright part of the pixel charge beyond the full trap ability, overflow to the surrounding neighboring pixels, the screen produces white patches.
28. Smear (Light leakage)
    • Smear phenomenon refers to when shooting the image of high brightness in the light or sunlight, high-brightness part of the upper and lower side can see very bright vertical white line or Leucorrhea. Different CCD structure, the mechanism of producing smear is also different.
30. Basic principles of camera selection
    • According to the requirements of the testing project and the characteristics of the target to be detected, select a surface array or a line array camera
    • Determine whether a color or monochrome camera is selected. In some color classification occasions, the use of monochrome cameras with different spectral sources or filters, will also make processing simplified
    • Defines the minimum feature and detection field of the target to be detected, thereby determining the camera's resolution
    • Determine the camera frame rate (line frequency) based on the target motion speed and pass rate
    • According to the processing demand, select the intelligent features of the camera, such as: Flat field correction, LUT, binning, etc.

How to select an industrial camera

Industrial cameras are an important part of machine vision systems and industrial inspection applications, but how do you choose the right industrial camera? Select industrial cameras, first of all to clear their own testing tasks, is the static or dynamic, the frequency of the camera, is to do defect detection or dimensional measurement, or positioning, the size of the product (field of view) is how much, the need to achieve how much precision, on-site environmental conditions, there are no other special requirements . Here is a summary based on our experience. Want to help everyone in the purchase of industrial cameras can do a good idea, really can choose to fit their own camera.

Analog Cameras & Digital Cameras
The analog camera must be digitally captured with a digital capture card and converted to a digital signal for transmission and storage. The general analog camera resolution is very low, the other frame rate is also fixed. analog signals can cause distortion due to electromagnetic interference from other equipment in the plant, such as electric motors or high-voltage cables . As the noise level increases, the dynamic range of the analog camera (the ratio of the original signal to the noise) decreases. Dynamic range determines how much information can be transferred from the camera to the computer.
Digital Camera Digital signals are captured and digital signals are not affected by electrical noise, so the digital camera has a higher dynamic range and is capable of transmitting more accurate signals to the computer. This depends on the actual needs to choose.

resolution
       resolution is not as high as possible, is to choose the size of the resolution according to the actual application needs of the system.
      application case: Assume that the detection of an object's surface scratches, required to shoot the size of the object is 10*8mm, the required detection accuracy is 0.01mm. First, assuming that we want to take the field of view in 12*10mm, then the minimum resolution of the camera should be selected in: (12/0.01) * (10/0.01) =1200*1000, about 1.2 million pixels of the camera, that is, a pixel corresponding to a detection defect, The minimum resolution must be no less than 1.2 million pixels, but a 1.3 million-megapixel camera is common on the market, so it is generally a 1.3 million-megapixel camera. But the real problem is that if a pixel corresponds to a flaw, then such a system will be extremely unstable, because a random interference pixel can be mistaken for defects, so in order to improve the accuracy and stability of the system, it is best to take the defect area of 3 to 4 pixels above, So we choose the camera is 1.3 million times 3 or more, that is, the minimum must not be less than 3 million pixels, usually using a 3 million-megapixel camera for the best (I have seen the most people holding sub-pixel to do fraction sub-pixel, then do not use such a high-resolution camera.) For example, they say that if you do 0.1 pixels, is a defect corresponding to 0.1 pixels, the size of the defect is calculated by the number of pixels, how to say 0.1 pixels of the area? These people use Subpixel to bluff people, often explain their lack of common sense). In other words, we are only used for the measurement, then the sub-pixel algorithm, the 1.3 million-megapixel camera can basically meet the demand, but sometimes because of the effect of edge sharpness, when extracting edges, arbitrarily offset a pixel, then the accuracy is greatly affected. Therefore, we choose 3 million of the camera, you can also allow the extraction of the edge deviation of 3 pixels, which is very good to ensure the accuracy of the measurement.

Ccd&cmos
if the object to be photographed is moving, the objects to be dealt with are real-time moving objects, of course, it is most suitable to choose the CCD chip camera with full frame exposure of Dee-Jie. However, some manufacturers of CMOS cameras if the use of frame exposure, it can also be used as a CCD. If the object movement is very slow, in our set camera exposure time range, the object movement distance is very small, conversion to pixel size is within one or two pixels, then the choice of CMOS camera is also appropriate. Because in the exposure time, one or two pixels of the deviation of the human eye can not see (if not to do the measurement), but more than 2 pixels deviation, the object shot out of the image there is a drag, so you can not choose a CMOS camera.

Color & Monochrome
if we are to deal with the color of the image, it is of course the use of color cameras, otherwise it is recommended that you use black and white, because the same resolution of black and white camera, precision than color, especially when looking at the edge of the image, black and white effect is better.

Frame rate
depending on the speed to be detected, select the camera frame rate must be greater than or equal to the detection speed, equal to the situation is that you process the image time must be fast, must be in the camera exposure and transmission of time to complete.

Line Array & surface array
For the detection accuracy requirements are high, the resolution of the face array camera is not up to the requirements of the case, of course, the line array camera is an inevitable choice.

Transport Interface
Select the camera for USB, 1394, Camerlink, Gigabit/Gigabit interface based on distance, stability, transmitted data size (bandwidth).

CCD target Surface
The size of the target size will affect the length of the lens focal length, in the same perspective, the larger the size of the target, the longer the focal length. When selecting the camera, especially when the shooting angle is more stringent, the CCD target size, CCD and lens with the situation will directly affect the size of the field of view and image clarity. Therefore, in the selection of CCD size, the lens should be combined with the focal length, the angle of view together to choose, in general, the choice of CCD target to be combined with the physical installation of space to determine whether the lens working distance within the installation space,the size of the lens must be greater than or equal to the camera's target size.

Choice of Lens
Select the lens focal length according to the environment, then select the lens angle according to the viewing range;

The price of the camera
The same parameters of the camera, different manufacturers prices vary, this depends on everyone with the manufacturers to communicate and negotiate. Generally speaking, if you have a quantity, the overall price is very different from the price you buy a single.

Examples of Industrial camera selection principles: such as the detection task is the size measurement, the product size is 18mm*10mm, the accuracy is 0.01mm, assembly line operation, the detection speed is 10 pieces/second, the field environment is the general industrial environment, do not consider interference problems. First know is the assembly line work, faster, so the use of progressive scanning camera, the field of view size we can be set to 20mm*12mm (considering the error of each mechanical positioning, the field of view than the object to enlarge properly), if you can get a good image (such as can play backlight), and our software measurement accuracy can be considered 1 /2 subpixel accuracy, then we need the camera resolution is 20/0.01/2=1000pixcel (pixels), the other direction is 12/0.01/2=600pixcel, that is, our camera resolution needs at least 1000*600pixcel, frame rate of 10 frames/ seconds, so the choice of 1024x768 pixels (software performance and mechanical accuracy can also be considered 1280*1024pixcel), frame rate of 10 frames per second or more.

Common technical parameters of digital camera

1. Resolution (Resolution)
The so-called resolution refers to the resolution of the screen, the camera every time the pixel points of the image Acquisition (Pixels), for the camera is generally directly related to the sensor cell number. The resolution that is commonly seen is expressed in multiplicative form, such as 1024x768, where 1024 indicates the number of points displayed horizontally on the screen, and 768 indicates the number of vertical points. For the analog camera is based on the video format, PAL system for the 768*576,NTSC system for 640*480.
2. Maximum frame rate/row frequency (line rates):
The camera captures the rate at which the image is transmitted, for the number of frames per second (frames/sec.) that the array camera typically collects, and for the number of rows (Hz) that the line array camera collects per second.
3. Exposure mode (Exposure) and shutter speed (Shutter)
For the line array camera is a progressive exposure, you can choose the fixed line frequency and external trigger synchronization of the acquisition mode, the exposure time can be consistent with the line period, can also set a fixed time; there are several common ways of frame exposure, field exposure and rolling line exposure, and digital cameras generally provide the function of external triggering drawing. Shutter speeds are typically up to 10 microseconds, and high-speed cameras can be faster.
4, pixel depth (Pixel Depth):
That is, the number of bits per pixel of data, generally used is 8Bit, for digital cameras will generally have 10Bit, 12Bit and so on.
5. Cell size (Pixel size):
The cell size and number of cells (resolution) together determine the size of the camera target. At present, the digital camera cell size is generally 3μm-10μm.
6. Spectral response characteristics (spectral Range):
Refers to the sensitivity of the cell sensor to different light waves, the general response range is 350nm-1000nm, some cameras in front of the target with a filter, filter out the infrared light, if the system needs to be sensitive to the infrared can be removed from the filter.
7, photosensitive area target size:
The camera sensor size, 1 inches, 2/3 inches, 1/2 inches, 1/3 inches, 1/4 inches and so on, the smaller the number of inches, the smaller the range of field of view.

The determination of the mode of transmission of the camera, the advantages and disadvantages of the current camera transmission mode and its application are described below:

1) Analog Camera (PCI capture card), the speed requirements are not high choice. Its advantages: stable, cost-effective; disadvantage: low frame rate, generally can only reach 25 frames-30 frames;

2) USB interface camera, the system only use a single camera can be selected, the requirements of high-speed can be selected first. Advantages: Do not need to occupy PCI slots, high frame rate, cost-effective; disadvantage: Account for the system CPU;

3) 1394-interface camera, the system can be used to a number of cameras when the first choice, high-speed when the need for the first choice. Advantages: does not account for the system CPU, high frame rate; Disadvantage: PCI slots, expensive. Speed and exposure when choosing an industrial digital camera, the speed of an object's imaging must be well considered. For example, suppose that, in the course of filming, an object is not moving during exposure and can be used as a relatively simple and inexpensive industrial camera; for stationary or slowly moving objects, the array industrial camera is best suited for imaging stationary or slow moving objects. Because the entire area of the array must be exposed at once, any movement in the exposure time will result in blurred images, but motion blur can be controlled by reducing exposure time or by using a flash; for fast moving objects, when using a faceted industrial camera for moving objects, consider the number of moving objects in an industrial camera during exposure time, and the smallest feature that can be represented by a pixel on an object, that is, the object resolution, The thumb rule in capturing an image of a moving object is that exposure must occur within a time when the object is moving less than one pixel. If the object you are collecting is moving at a constant speed of 1 centimeters per second, and the object resolution is set to 1 pixel/mm, the maximum exposure time required is 1/10 per second. Because the object moves a distance exactly equal to one pixel in the camera sensor, there is a certain amount of blur when the maximum exposure time is used. In this case, it is generally preferred to set the exposure time above the maximum value, such as 1/20 per second, to keep the object moving half a pixel in the image. If the same object moves at 1 centimeters per second and the object resolution is 1 pixel/microns, the maximum exposure required in a second is 1/10000. The pair of exposure settings depends on the camera used, and whether you can give the object enough light to get a good picture.   choose a digital or analog camera

The
1. image processor and capture card are relatively easy-to-choose electronic devices whose main parameters are storage capacity and processing speed.
2. CCD and CMOS both have no obvious advantages and disadvantages in image quality. CMOS-based industrial cameras require fewer components, lower power consumption, and faster data than CCD-based cameras, but the CCD is a more sophisticated technology capable of providing better quality images with lower noise, while the weaknesses are slower, less flexible data transfers, More parts and higher power consumption .
3. affect camera price factor
   • Typically, the higher the number of pixels, the more expensive the camera is due to sensor size.
   • at a certain resolution, the frame rate increases and the cost tends to increase. Increasing frame rate and resolution at
   • often requires the camera to have multi-port readout, which increases the complexity of the system and thus increases the cost. The
4. CCD and CMOS chip generate analog signals internally, so the main difference between the analog camera and the digital camera is where the image is digitized. Digital cameras digitize the signal in the camera and transmit the signal digitally to the computer (or image processor) via a serial bus interface (such as FireWire, USB, Camera Link, Gigabit Ethernet). On the other hand, the analog camera system does not digitize the image signal internally (digitized is done by the computer), so the analog information is transmitted through the coaxial cable.
5. Although both methods can effectively transmit the signal, the analog signal may cause distortion due to electromagnetic interference from other equipment in the plant, such as electric motors or high-voltage cables. As the noise level increases, the dynamic range of the analog camera (the ratio of the original signal to the noise) decreases. The dynamic range determines how much information can be transferred from the camera to the computer.
6. Typical analog cameras require 5 watts to 10 watts of operating power, while digital cameras with comparable resolution indicators are less than 1 Watts
7. Conclusion: The performance and cost of industrial cameras should be carefully considered when selecting cameras for machine vision systems. Although industrial analog cameras are far cheaper than industrial digital cameras, their resolution and image quality are low, so they may be confined to applications that are not very demanding. Digital cameras are more expensive than analog cameras, but their high costs may be worth the cost of demanding high-speed, high-accuracy, and high-precision applications.

Machine Vision-Camera