US-12626348-B2 - Analysis apparatus, inspection system, and learning apparatus

US12626348B2US 12626348 B2US12626348 B2US 12626348B2US-12626348-B2

Abstract

An analysis apparatus includes: a hardware processor that: acquires image information items of a plurality of images regarding a target that are captured while the target is irradiated with light, extracts, based on the image information items, an image in which an irradiation region where the target is irradiated with the light and an inspection target region of the target have a predetermined relationship, from among the images, and analyzes a state of the inspection target region based on each of the image information items of the extracted image.

Inventors

Takehiko Sashida

Assignees

Konica Minolta, Inc.

Dates

Publication Date: 20260512
Application Date: 20211118
Priority Date: 20201130

Claims (15)

1 . An analysis apparatus comprising: a hardware processor that: acquires image information items of images regarding a target that are captured while the target is irradiated with light, extracts, based on the image information items, an image in which a peripheral portion of an irradiation region of the target overlaps at least a part of an inspection target region of the target, from among the images, wherein the irradiation region is a region irradiated with the light, and includes: a central portion that has uniform luminance; and the peripheral portion that is outside of the central portion and has non-uniform luminance, and analyzes a state of the inspection target region based on each of the image information items of the extracted image.
2 . The analysis apparatus according to claim 1 , wherein the hardware processor extracts an image in which a peripheral edge of the irradiation region overlaps the inspection target region, from among the images.
3 . The analysis apparatus according to claim 1 , wherein the hardware processor extracts an image in which the inspection target region has non-uniform luminance, from among the images.
4 . The analysis apparatus according to claim 1 , wherein the hardware processor extracts the image based on a difference between a maximum luminance and a minimum luminance of the inspection target region and an average luminance of the inspection target region.
5 . The analysis apparatus according to claim 1 , wherein the hardware processor extracts the image based on at least one of a distribution and a histogram of luminance of the inspection target region.
6 . The analysis apparatus according to claim 1 , wherein the hardware processor acquires the image information items of the images regarding the target that are captured while any one of the target, the irradiation region, and an image capturing position moves.
7 . The analysis apparatus according to claim 6 , wherein the hardware processor further: identifies the inspection target region in each of the images based on each of the image information items, tracks the inspection target region in each of the images based on the identified inspection target region, and extracts the image based on the tracked inspection target region.
8 . The analysis apparatus according to claim 1 , wherein the hardware processor further: extracts a plurality of the images in each of which the peripheral portion overlaps at least a part of the inspection target region, from among the images, and analyzes the state of the inspection target region based on the image information items of the images.
9 . The analysis apparatus according to claim 1 , wherein the hardware processor analyzes the state of the inspection target region using a learned model.
10 . The analysis apparatus according to claim 9 , wherein the learned model is learned in advance by using training data of a combination of the inspection target region in the extracted image and a ground truth label of the state of the inspection target region.
11 . The analysis apparatus according to claim 1 , wherein the hardware processor analyzes the state of the inspection target region using deep learning.
12 . The analysis apparatus according to claim 1 , wherein the inspection target region is a candidate region of a defect in the target, and the hardware processor analyzes a shape of the defect.
13 . The analysis apparatus according to claim 12 , wherein the shape is a recessed shape or a protrusion shape.
14 . An inspection system comprising: a light source apparatus that irradiates a target with light; an imaging apparatus that images the target irradiated with the light from the light source apparatus; and the analysis apparatus according to claim 1 .
15 . A learning apparatus comprising: a hardware processor that: acquires image information items of a plurality of images regarding a target that are captured while the target is irradiated with light, extracts, based on the image information items, an image in which a peripheral portion of an irradiation region of the target overlaps at least a part of an inspection target region of the target, from among the images, wherein the irradiation region is a region irradiated with the light, and includes: a central portion that has uniform luminance; and the peripheral portion that is outside of the central portion and has non-uniform luminance, analyzes a state of the inspection target region based on each of the image information items of the image extracted using a learned model, and causes the learned model to perform further learning.

Description

CROSS-REFERENCE TO RELATED APPLICATION The entire disclosure of Japanese Patent Application No. 2020-198500 filed on Nov. 30, 2020, including description, claims, drawings, and abstract, is incorporated herein by reference. BACKGROUND Technical Field The present invention relates to an analysis apparatus, an inspection system, and a learning apparatus. Description of Related Arts Defects may occur on a coated surface of an automobile or the like. Development of an inspection system for detecting such a defect is underway (e.g., Patent Literature 1). In this inspection system, for example, a defect is detected by sequentially imaging a partial region of an automobile or the like under light irradiation. PATENT LITERATURE Patent Literature 1: JP 2000-172845 A It is difficult to improve accuracy of an inspection in an inspection system for detecting a defect in a coated surface of an automobile or the like. SUMMARY One or more embodiments of the present invention provide an analysis apparatus, an inspection system, and a learning apparatus that can improve the accuracy of the inspection or detecting a defect in a coated surface of an automobile or the like. One or more embodiments of the present invention deal with the above issues by the following means. (1) An analysis apparatus including: a hardware processor that: acquires image information items of images regarding a target that are captured while the target is irradiated with light; extracts, based on the image information items, an image in which an irradiation region where the target is irradiated with the light and an inspection target region of the target have a predetermined relationship, from among the images; and analyzes a state of the inspection target region based on the image information items of the extracted image. (2) The analysis apparatus according to (1), wherein the irradiation region has a central portion and a peripheral portion outside the central portion, and the hardware processor extracts an image in which the peripheral portion of the irradiation region overlaps at least a part of the inspection target region, from among the images. (3) The analysis apparatus according to (1), wherein the hardware processor extracts an image in which a peripheral edge of the irradiation region overlaps the inspection target region, from among the images. (4) The analysis apparatus according to (1), wherein the hardware processor extracts an image in which luminance of the inspection target region has predetermined non-uniformity, from among the images. (5) The analysis apparatus according to (1), wherein the hardware processor extracts the image based on a difference between a maximum luminance and a minimum luminance of the inspection target region and an average luminance. (6) The analysis apparatus according to (1), wherein the hardware processor extracts the image based on at least one of a distribution and a histogram of luminance of the inspection target region. (7) The analysis apparatus according to (1), wherein the hardware processor acquires the image information items of the images regarding the target that are captured while any one of the target, the irradiation region, and an image capturing position moves. (8) The analysis apparatus according to (7), wherein the hardware processor identifies the inspection target region in the images based on the acquired image information items and tracks the inspection target region in the images based on the identified inspection target region, and extracts the image based on the tracked inspection target region. (9) The analysis apparatus according to (1), wherein the hardware processor extracts a plurality of the images in each of which the irradiation region and the inspection target region have a predetermined relationship, from among the images, and analyzes the state of the inspection target region based on the image information items of the plurality of the images. (10) The analysis apparatus according to (1), wherein the hardware processor analyzes the state of the inspection target region using a learned model. (11) The analysis apparatus according to (10), wherein the learned model is learned in advance by using training data of a combination of the inspection target region in the image extracted and a ground truth label of the state of the inspection target region. (12) The analysis apparatus according to (1), wherein the hardware processor analyzes the state of the inspection target region using deep learning. (13) The analysis apparatus according to any one of (1), wherein the inspection target region is a candidate region of a defect in the target, and the hardware processor analyzes a shape of the defect. (14) The analysis apparatus according to (13), wherein the shape is a recessed shape or a protrusion shape. (15) An inspection system including: a light source apparatus that irradiates a target with light; an imaging apparatus that images the target irradiated with light from th