Defect inspection method, defect inspection device and defect inspection system
An optical inspection method for an optical inspection device comprising an optical lens is provided according to an embodiment of the disclosure. The optical inspection method includes: obtaining a first image of an object by the optical lens; performing an edge detection on the first image to obtain a second image comprising an edge pattern; and performing a defect inspection operation on the second image based on a neural network architecture to inspect a defect pattern in the second image. In addition, an optical inspection device and an optical inspection system are provided according to embodiments of the disclosure.
1. A defect inspection method for an optical inspection device with an optical lens, wherein the defect inspection method comprises:
obtaining a sample image;
generating a first non-defect image according to the sample image and adding a first defect pattern into a first image region randomly, so as to generate a first defect image according to the sample image,
wherein the first defect image comprises the first defect pattern, the first non-defect image does not comprise the first defect pattern, and the first non-defect image and the first defect image both correspond to the first image region of the sample image;
training a neural network architecture according to the first non-defect image and the first defect image;
obtaining a first image of an object by the optical lens;
performing an edge detection on the first image to obtain a second image with an edge pattern; and
performing a defect inspection operation on the second image with the edge pattern based on the trained neural network architecture to inspect a defect pattern in the second image.
2. The defect inspection method according to claim 1 , wherein the step of performing the edge detection on the first image to obtain the second image comprising the edge pattern comprises:
performing the edge detection on the first image based on at least one of a red model, a green model and a blue model of a RGB color model, so as to obtain the second image comprising the edge pattern.
3. The defect inspection method according to claim 1 , wherein the step of generating the first non-defect image according to the sample image and adding the first defect pattern into the first image region randomly, so as to generate the first defect image according to the sample image comprises:
performing at least one of a chromaticity adjustment operation, a brightness adjustment operation and a pattern scaling operation, so as to generate the first non-defect image and the first defect image.
4. The defect inspection method according to claim 1 , wherein the step of adding the first defect pattern into the first image region randomly comprises:
selecting at least one first pixel in the first image region;
selecting at least one second pixel neighboring to at least one of the at least one first pixel in the first image region; and
generating the first defect pattern according to the at least one first pixel and the at least one second pixel.
5. A defect inspection device, comprising:
an optical lens, configured to obtain a first image of an object; and
a processing circuit, coupled to the optical lens and configured to:
obtain a sample image,
generate a first non-defect image according to the sample image and add a first defect pattern into a first image region randomly, so as to generate a first defect image according to the sample image,
wherein the first defect image comprises the first defect pattern, the first non-defect image does not comprise the first defect pattern, and the first non-defect image and the first defect image both correspond to the first image region of the sample image,
train a neural network architecture according to the first non-defect image and the first defect image,
perform an edge detection on the first image to obtain a second image comprising an edge pattern,
perform a defect inspection operation on the second image with the edge pattern based on the trained neural network architecture to inspect a defect pattern in the second image.
6. The defect inspection device according to claim 5 , wherein the operation of performing the edge detection on the first image to obtain the second image comprising the edge pattern comprises:
performing the edge detection on the first image based on at least one of a red model, a green model and a blue model of a RGB color model, so as to obtain the second image comprising the edge pattern.
7. The defect inspection device according to claim 5 , wherein the operation of generating the first non-defect image according to the sample image and adding the first defect pattern into the first image region randomly, so as to generate the first defect image according to the sample image comprises:
performing at least one of a chromaticity adjustment operation, a brightness adjustment operation and a pattern scaling operation, so as to generate the first non-defect image and the first defect image.
8. The defect inspection device according to claim 5 , wherein the operation of adding the first defect pattern into the first image region randomly comprises:
selecting at least one first pixel in the first image region;
selecting at least one second pixel neighboring to at least one of the at least one first pixel in the first image region; and
generating the first defect pattern according to the at least one first pixel and the at least one second pixel.
9. A defect inspection system, comprising:
an optical inspection device, configured to obtain a first image of an object and perform an edge detection on the first image to obtain a second image with an edge pattern; and
a server, coupled to the optical inspection device and configured to:
obtain a sample image,
generate a first non-defect image according to the sample image and add a first defect pattern into a first image region randomly, so as to generate a first defect image according to the sample image,
wherein the first defect image comprises the first defect pattern, the first non-defect image does not comprise the first defect pattern, and the first non-defect image and the first defect image both correspond to the first image region of the sample image,
train a neural network architecture according to the first non-defect image and the first defect image, and
perform a defect inspection operation on the second image with the edge pattern based on the trained neural network architecture to inspect a defect pattern in the second image.
10. The defect inspection system according to claim 9 , wherein the operation of performing the edge detection on the first image to obtain the second image comprising the edge pattern comprises:
performing the edge detection on the first image based on at least one of a red model, a green model and a blue model of a RGB color model, so as to obtain the second image comprising the edge pattern.
11. The defect inspection system according to claim 9 , wherein the operation of generating the first non-defect image according to the sample image and adding the first defect pattern into the first image region randomly, so as to generate the first defect image according to the sample image comprises:
performing at least one of a chromaticity adjustment operation, a brightness adjustment operation and a pattern scaling operation, so as to generate the first non-defect image and the first defect image.
12. The defect inspection system according to claim 9 , wherein the operation of adding the first defect pattern into the first image region randomly comprises:
selecting at least one first pixel in the first image region;
selecting at least one second pixel neighboring to at least one of the at least one first pixel in the first image region; and
generating the first defect pattern according to the at least one first pixel and the at least one second pixel.