Defect image classification collection method based on VAI surface detection system

Abstract

The invention belongs to the strip steel surface detection technical field, and discloses a defect image classification collection method based on a VAI surface detection system; the method comprisesthe following steps: based on a first attribute characteristic constant of defect images, screening a first defect image group satisfying the first attribute characteristic constant scope from to-be-classified images; based on a second attribute characteristic constant of the defect images, and screening initial classification defect images satisfying the second attribute characteristic constant scope from the first defect image group, wherein the first attribute characteristic constant includes a defect area, a defect length and a defect width; the first attribute characteristic constant scope comprises a defect area scope, a defect length scope and a defect width scope; the second attribute characteristic constant comprises an average grey value and a grey value standard deviation; thesecond attribute characteristic constant scope comprises an average grey value scope and a grey value standard deviation scope; the first and second attribute characteristic constant scopes are determined according to specific defect types. The invention provides the reliable classification collection defect image method.

Description

technical field [0001] The invention relates to the technical field of steel strip surface detection, in particular to a method for classifying and collecting defect pictures based on a VAI surface detection system. Background technique [0002] The classification and collection of defect pictures of the VAI linear array surface inspection system (hereinafter referred to as VAI) is an important preparation process for the establishment of the VAI defect database. Mainly by looking for the corresponding defect pictures for various known defects as the original defect data for establishing the database. The results of VAI defect classification and collection directly affect the generation of spherical files and the classification effect of the classifier. Therefore, we must ensure that the defect images are qualitatively accurate when we classify and collect VAI defect images. [0003] However, in the prior art, when collecting defect pictures of VAI classifiers, the shape is...

AI Technical Summary

Problems solved by technology

[0004] The invention provides a method for classifying and collecting defect pictures based on a VAI surface detection system, which solves the technical problem that the qualitative classification of the original defect pictures in the defect database of the VAI linear array surface detection system is subject to subjective influence and poor reliability in the prior art