What should companies pay attention to before implementing an automatic visual

        practical automated visual inspection system can significantly reduce the risk of defective products and reduce production costs over time. Before implementing an automated visual inspection system project, companies need to pay attention to the following factors:

1. Each part has its own solution

Designing automated visual inspection systems to test different part types has proven to be a difficult task as part geometry may be occluded or shadows may hide relevant areas. These limitations are often a result of product design because automatic detection is not part of the product design process. It is through illumination and processing cycle time defined defect specifications, illumination, resolution and camera speed that help obtain accurate analysis.

If there is a suspected defect, the operator has a chance to take another look, but the machine does not. During the design process of an automated visual inspection system (AVC), possible defects in each product must be considered. Even similar parts have challenges associated with specific materials or product designs, so finding ways to customize them is necessary.

Automated visual inspection systems are complex combinations of cameras, lighting and processing systems, including some complex image processing

If a company is looking into implementing an automated visual inspection system for quality inspection, it makes the most business sense to start with the company's highest volume parts or very similar parts such as O-rings. O-ring inspection systems can inspect a variety of different components. Configuring the system to test different types of O-rings is less difficult because O-rings are very similar to each other and have simple geometries; therefore, different part types can be inspected using a single system.

2. Dimensional measurement and surface inspection

There are two types of automatic visual inspection systems in general factories. The first type is mainly used for dimensional measurement, and the second type is used for surface defect detection. Of the two, the dimensional measurement system is the easiest to develop. Furthermore, statistical methods for calculating solution capabilities (e.g., measurement system analysis) are easy to design for machines because everything about an automated visual inspection system can be estimated in advance. By clearly defining part specifications, the system can identify critical parameters, even if part dimensions have the tightest tolerances. Once these critical parameters are confirmed, the correct implementation of the inspection system can begin, using computational methods to verify relevant parameters such as system lighting, speed or camera resolution.

On the other hand, surface inspection can be very challenging. Having documented instructions that tell human quality inspectors what defects and dimensions to check for when manually inspecting parts facilitates the development of automated visual inspection systems.

However, these instructions do not provide sufficient guidance for software engineers developing programs for automated visual inspection systems. While human inspectors may fully understand these instructions, software developers need access to more information.

For example, a quality directive may declare a specific type of defect NG. Taken literally, this would require an inspection system with infinite resolution. Rather than simply saying NG, developers must provide quantifiable defects so that they can fit into the system design requirements. If very small defects are NG, the system needs to be able to identify them, thus requiring higher image resolution.

Companies often face two challenges: A defect needs to occur and be noticed before it can be corrected; and communicating the characteristics of a specific defect to a computer is difficult. Therefore, companies usually implement "basic" inspections in all areas to at least have a chance of discovering unknown defects. Traditionally, machines have been built with these flaws in mind because they were already available. Some defect predictions may be possible, but increase the risk of “pseudo” rejects.

Reaching agreement on the size of a defect can be difficult because many factors must be taken into consideration. Quality engineers must work with design engineers and even machine suppliers to come up with the right metrics.

3. Definition and classification of defects

Because defining and classifying defects is a very important part of the development process, quality engineers should compile a defect catalog before beginning automated visual inspection system development. A defect classification is not only a list of all defects that the system must be able to detect, but also a collection of those parts that are within or close to the acceptable range. The reason why this part is so important is that the system will be specifically verified by checking these parts.

As shown in the spreadsheet, the defect catalog includes the type of defect, its severity, probability, and critical size of the defect. All this information helps software developers prioritize detection of common defects.

Defect definition classification is the most critical prerequisite for successful automated visual inspection system project implementation.

For each defect, it is best to have at least two samples, one of which is an edge defect. Depending on the complexity of the part, defect classifications typically range from 60 to 100 parts. It is usually best to have enough samples, as it is easier to identify parts that do not add knowledge to the catalog than to discover that important information is missing.

The biggest challenge in developing a defect catalog classification is communication between the quality department and the visual inspection system developers. Visual inspection system developers need to understand the production process, and quality departments must understand the limitations of the inspection system.

For example, quality engineers understand some of the terminology used in manufacturing processes. However, specific characteristics such as shape, softness and thickness must be described in detail to vision developers to ensure the system can find defective parts rather than reject good ones. Depending on the cause and type of defect, different methods are required to detect these defects.

4. System Check Process

In order to reliably find a specific defect, the program must perform five steps:

1. Part positioning: The program must position the part in the image to compensate for small changes in the processing system.

2. Image segmentation: The program segments the image into functional areas of interest. Each zone is defined by a limited set of defects and a single feature relative to the part, such as sealing lip, outer diameter, top surface, etc.

3. Image normalization: Remove from the image everything that is usually found in the good parts. This includes design/feature lines and common product variations. This is a core feature because it compares each new image to the history of good parts and highlights deviations from this history.

4. Signal/noise optimization: At this point, the program filters the image to extract possible defects from known noise. This step requires consideration of specific defect characteristics such as dark contamination, bright marks, and vertical flow lines.

5. Defect detection and classification: Finally, the program identifies suspicious areas and assigns a classification value, which is used in pass or reject decisions. In addition, the program records whether the part has acceptable defects. This data can be used to prevent production problems before they occur.

5. Ensure high-quality images

Good inspection relies on high-quality images. Here, there are several things a factory can do to ensure that the visual inspection system produces the best possible image. Image quality often depends on lighting and camera settings, but defect types must also be considered. Other factors include reducing the number of false rejections; staying within required cycle times; and making full use of available machine space.

Optimal settings allow inspection of the smallest relevant defects

For example, a shop can take images of tools before and after cleaning or of parts made with different tool sets to see how process changes will affect image quality. By analyzing these images and applying its knowledge of the limitations of image processing software, the company will be able to come up with a workable solution. Software limitations sometimes mean companies have to develop more complex lighting and camera setups. Other times, hardware limitations increase program complexity or development time, or in some cases, mean you have to tolerate more bug rejections than you'd like.

6. Big investment = big return

Developing and installing an automated visual inspection system is always a significant investment for any company, but it’s an investment that often pays off. The biggest benefit is reduced costs, because automatic inspection costs less than manual inspection. Another benefit is that developing a custom automated visual inspection system can significantly reduce the risk of performing poor inspections and delivering defective parts. It can also detect anomalies that have not yet manifested defects. This data can alert engineers to problems so they can correct them before defective parts are produced.

Automated visual inspection systems reduce the number of customer complaints, and for some part types the complaint rate can be reduced to zero. This trend is likely to continue to grow as "intelligent systems" train themselves to find parts that differ from typical products. Smart systems can make automated inspection systems more attractive because they reduce the risk of programming errors, as well as the time and cost of implementing new solutions.