Significance of FlexInspector in the Imaging and Inspection Industry

We believe that FlexInspector's inspection method will bring about a major revolution in the image inspection industry.
1) "Binarization Level per Pixel
Even a low-resolution camera has an upper brightness limit and a lower brightness limit (binarization level) for each of the 300,000 pixels that exist. This simple method does not slow down the speed and opens up the possibility of image inspection of objects with a gradient in brightness, especially three-dimensional objects such as moldings. This method is unique in the world.
(2) Statistical processing of variation in good products
One of the problems in "good part comparison" such as pattern matching and differential processing is "variation of good parts. FlexInspector statistically processes multiple images of good parts and absorbs the variation of good parts. As a result, FlexInspector is automatically tuned to the optimum level (yield and detection capability) according to the "variation of good parts," resulting in "almost parameter-free" image inspection.
(3) "Non-recognition" image inspection
Image inspection up to now has been achieved by extracting (i.e., recognizing) "where," "how," and "what" information from the image and setting "what it should be like. The problems with this method are
The information extracted is only a simple "numerical value" or "character string," which is only a "small part" of the information that is representative of the original image. The question is whether it is possible to make a judgment based on this "very small portion" of information.
In order to judge the entire image, it is necessary to repeat the above settings indefinitely. This is the main cause of the "difficulty" and "complexity" of the settings, and also the main cause of the inability to solve the problem of "vulnerability to unexpected defects" no matter how many settings are made.
Etc. It is true that "recognition" may be interesting as a study. However, in the field of product inspection, there should be no problem if a "good product" exists and if it can be mechanically determined whether it is the same as a "good product" or not, and if it can be sorted as good or bad. If "good products" cannot be precisely defined, we have no choice but to rely on "recognition," but experience shows that this is rarely the case.
(4) Making the system "usable" by on-site workers
Until now, image processing inspections were difficult and cumbersome, requiring several hours of setup by dedicated personnel. We have reviewed the fundamental concept of this system and made it usable. Even if the object changes, the setup procedure is always the same. Many items can be put into operation in a few minutes. Furthermore, it has numerous features: it only allows good products to pass through, it is resistant to unexpected defects, it can understand how it is being inspected, and it is cost-effective because it runs on an ordinary PC.
Perhaps it is even possible to make the "image inspection systems" that exist and are sold today easier, more accurate, and less expensive by incorporating FlexInspector. Furthermore, it is possible to build systems in fields that did not exist until now.
Also, since it is "simple" and "general-purpose," people in the field are starting to "look for places where they can use it themselves. I think this approach is the best way to "not let the image processing inspection theme fail.
This movement is certainly spreading.
There is some fear of what this movement will bring to the industry when it begins to accelerate.