As the expert in appearance inspection and image processing inspection, we have been well received for our high inspection accuracy, and our inspection systems have been installed in many industries and manufacturing processes.

The inspection accuracy is not just a matter of superior defect detection capability, but another important point is that the inspection algorithm incorporates measures to prevent matters like over-detection, missed defects, and other common problems that occur in manufacturing, such as a defect that can be observed by human eyes but not reflected in the image. Among these, we will introduce the irradiation technology using multiple lights, which is particularly important.

Why do we need multiple lights in image inspection?

In conventional image inspection, it is common to perform inspection using a single light. In contrast, ViSCO proposes to install multiple lights when there are multiple inspection items or combinations of different materials, as shown below.

By irradiating each target inspection item with the optimal lighting conditions, it is possible to acquire images showing defects clearly.

If this is done with a single light, it may result in capturing an image with unclear differences from the background depending on the defect, as shown below, which may lead to missed detection or over-detection, resulting in a decrease in inspection accuracy.

As introduced in previous article , there is an element called "perception judgment" a part of ViSCO's "algorithm that inspects like a human". This is a way of re-creating the judgments that human visual inspectors make based on their experience, such as "it will be easier to see if I do this” in the inspection system.

Lighting technology is classified in this category and is installed based on a comprehensive judgment of different lighting colors, shapes and number of lights, angles of illumination, and irradiation distances, etc., and also the size, material, and other characteristics of the object. This is a particularly important element of ViSCO's overall image inspection system, as these creativities play a role to make it easier to see defects from captured images, and perform subsequent image processing inspections.

DefFinder®︎ with human-like learning capabilities

Images captured with these creativities are instantly transferred to the VTV-9000, image processing is performed, and inspection results are output. What helps in this process is our originally developed inspection tool, DefFinder®︎

DefFinder®︎ was developed based on the insight from voices from worksites that "Even good products have variations.” At the time of inspection settings, a "statistically correct non-defective product image" is generated by training multiple good product images with slightly different colors and sizes, instead of registering only a single image of a good product. The system then uses these images to make a difference from the captured images and stably extract only the defective parts.

The tests that can be performed using this method are wide-ranging and have various targets. The following is just a few examples.

In addition, it has been developed in a series as an inspection tool for VTV-9000 software, and various applications are possible.

We have set up a YouTube channel with videos in English and Thai versions so that more customers can understand about image processing inspection and our company. You can also select Japanese with subtitles. Please subscribe to our channel and watch the videos.

For inquiries regarding automating the appearance inspection of industrial products in Thailand, please contact us using the inquiry form below.