Methods of improving machine vision system

With the increasingly fierce competition in the global manufacturing market , It is important for enterprises to ensure that they operate with the highest efficiency . Downtime due to process related failures is considered non value added downtime , Directly affect the profitability of the company . Machine vision is a new field , If the settings are correct , It can reduce the inefficiency of the process . Any facility that uses machine vision as robot guidance or inspection should pay attention to this field , To seek possible improvements . If there is no appropriate setting in the machine vision system , It may cause non value-added downtime of machine vision process , Here are ten areas that need special attention for improvement ：

1) Lighting technology - The correct lighting technology should be used to illuminate the area to be detected . Backlight 、 Bright field 、 grazing 、 Low angle linear array and dark field illumination technology are the most critical aspects of machine vision robustness . According to the surface finish and contour of the part , Correct lighting technology can enhance defects or remove image noise , Improve the efficiency and stability of the system . The purpose of this is to choose a method that can produce the maximum contrast ( From black to white pixels ) Lighting technology . Besides , Contrast needs to be directly related to what is being measured or checked .

2) Lighting color - The light color used for each specific part or application should be considered . The frequency is the number of oscillations per second , And the wavelength is the distance between two points on the same position . Each kind of different UV 、 blue 、 green 、 yellow 、 red 、 Infrared spectra have different illumination frequencies and wavelengths . These changes will affect the response of objects and camera surfaces when light enters . Its purpose is to use the optical frequency that can produce the maximum contrast and eliminate the noise in the image . for example , Metal parts can sometimes be introduced into a system , The system has a thin layer of oil or a slight oxidation on the surface , It depends on how they are stored . When both types of parts are introduced into the inspection system , It is important to use the frequency of light to reduce fluctuations .

3) Use a filter filter , Eliminate serious environmental interference such as background and overhead lighting noise . By simply placing a filter on the camera lens that matches the frequency of the light illuminating the component , It can eliminate the interference of ambient lighting .

4) The lens - field (FOV) And regions of interest (ROI), Including the required pixel accuracy , It plays an important role . The correct focal length lens will determine the size of the area that the machine vision system can see , And finally decide all the information collected . Calculation is too large FOV Will result in less detail and accuracy , And the calculation is too small FOV It may cause the inspection to fail , Because parts or objects are out of the line of sight of the camera . In the calculation FOV when , It is important to decide which focal length lens is best for the application , First determine the maximum of the part or object ROI And the maximum acceptable error in this area . Sometimes these factors may be limited by the working distance or height from the camera to the object , So all these factors need to be considered before building the system .

5) location —— It is important to know that the tolerance of a part or application is too wide , Repeated inspection of components is not allowed . When a component moves out of the camera's field of view , It will lead to system instability . Some type of physical fixture should be used to limit the movement of objects or parts of interest . If the part moves out of the camera view , It's going to break down , Increase unnecessary downtime . By providing a rough location of the part , This instability can be eliminated , Ensure that the parts can appear repeatedly in the machine vision system every time .

6) calibration - Have master fixture or calibration program , The system can be properly calibrated , Ensure that it meets the quality standards of the equipment .

Features and datums —— In order to check the image correctly , The benchmark —— The only feature found in each inspection —— It can be used as a reference point for visual tools during inspection , Or it can be used to detect whether there are correct parts in the image .

8) The resolution of the - The resolution determines the repeatability of the visual system ; It allows the size of pixels to be quantified into measured values . It is important to determine the resolution of the system , Because it determines the accuracy and repeatability of the inspection . In particular, solutions that need to be considered in quality measurement and detection and robot guidance . Some software can improve the resolution , The accuracy can reach sub-pixel .

9) stability —— When setting up the machine vision sensor , It is important to protect the camera system and light from moving . These items are the goals of system calibration . To minimize movement or interference , It is wise to place the camera and lighting lamp in a place with little or no vibration and traffic .

10) test —— Check the system regularly , Confirm whether the defective parts are captured and rejected by the system . You can build a test process directly in the system , To simplify the process . Defective test parts can be placed in the system at any time , To verify the proper function of the system .