There are serious consequences caused by machine vision vulnerabilities, such as image deviation. The following details how to check for machine vision vulnerabilities.

    1.Defect inspection based on statistical features such as texture and blob

统计 According to the surface gray distribution characteristics of the product, high-order statistical factors and edge detection operators are used to separate the defect points from the background to quickly identify the defect location.

    2. Defect inspection based on color features

The color characteristics of the surface of the object to be tested are statistically processed based on the vector characteristics of multiple color spaces such as RGB, YCbCr, HSI, and other methods such as pattern classification and feature learning to distinguish defect features from normal features and determine And identify defect locations. Mainly used in printing and packaging, PCB manufacturing, medical equipment and other fields.

    3. Defect inspection based on geometric features

Mainly for the product's geometric shape characteristics, combined with known feature information such as CAD, the corresponding area of the product is subjected to noise processing, feature extraction, fitting, parameter extraction, and then compared with the design features. The location of the area is determined.