The essential technological elements required for image processing determination include "Technologies Related to Image Acquisition (technologies that replace the "eyes" of visual determination)" and "Technologies Involved in Pass/Fail Classification (technologies that replace the "brain" of visual discrimination)," along with "Technologies related to data utilization."

Through image processing determination, various information is accumulated as digital data. In this section, we will explain what kind of data is accumulated and the effects obtained through effective utilization of the accumulated digital data.

1. Data Accumulated Through Image Processing Discrimination

It is possible to obtain various information such as the determination result (OK/NG), image information at the time of determination, details of the determination (items extracted as defects, quantity, size, position, etc.), inspection date and time, lot number, and inspection quantity.

2. The Utility of Data

The digital data obtained through image processing is a crucial element directly contributing to productivity enhancement and competitive strengthening.

Data analysis is expected to yield the following effects:

Yield Improvement: Analyzing data on the occurrence and location of images classified as defective, along with data acquired from production equipment and workflow processes, may lead to identifying factors causing defects.

Addressing these identified factors can result in yield improvement.

Traceability Assurance: Storing image data from image processing enables confirmation of product status via data.

This facilitates swift customer response and minimizes inspection scope in case of reinspection.

Early Detection of Defect Occurrence: With defects worsening over time, image processing allows grasping data to detect precursors to defects at an early stage, thus suppressing defect occurrence.

3. Summary

Correctly distinguishing between pass and fail products is the foundation of image processing classification.

However, solely using acquired data for this purpose would be squandering its potential.

By analyzing the acquired data and feeding back change points or trend values into the process, it becomes possible to further enhance the value of image processing classification.