Information processing apparatus and information processing method

Patent metadata

Field	Value
Publication number	US-11715194-B2
Application number	US-202117191063-A
Country	US
Kind code	B2
Filing date	Mar 3, 2021
Priority date	Sep 4, 2020
Publication date	Aug 1, 2023
Grant date	Aug 1, 2023

Abstract

An information processing apparatus has an acquisitor configured to acquire an entire area image obtained by capturing an entire area of a processing surface of a wafer including at least one defect, a training image selector configured to select, as a training image, a partial image including at least one defect from the entire area image, a model constructor configured to construct a calculation model of generating a label image obtained by extracting and binarizing the defect included in the partial image, and a learner configured to update a parameter of the calculation model based on a difference between the label image generated by inputting the training image to the calculation model and a reference label image obtained by extracting and binarizing the defect of the training image.

First claim

The invention claimed is:

1. An information processing apparatus comprising:
an acquisitor configured to acquire an entire area image obtained by capturing an entire area of a processing surface of a wafer including at least one defect;
a training image selector configured to select, as a training image, a partial image including at least one defect from the entire area image;
a model constructor configured to construct a calculation model of generating a label image obtained by extracting and binarizing the defect included in the partial image; and
a learner configured to update a parameter of the calculation model based on a difference between the label image generated by inputting the training image to the calculation model and a reference label image obtained by extracting and binarizing the defect of the training image.

2. The information processing apparatus according to claim 1 , wherein
the training image selector is configured to acquire, as a plurality of the training images, a plurality of the partial images in which pixel positions of at least a part in the entire area image are different and each of which includes at least one defect, and
the learner is configured to sequentially calculate differences between a plurality of the label images generated by sequentially inputting the plurality of training images to the calculation model and the corresponding reference label images, and sequentially update parameters of the calculation model.

3. The information processing apparatus according to claim 1 , wherein
the training image selector is configured to select, as the training image, the partial image included in an area selected from the entire area image.

4. The information processing apparatus according to claim 3 , further comprising:
an area selector configured to select the area from the entire area image, wherein
the training image selector is configured to select, as the training image, the partial image included in the area selected from the entire area image.

5. The information processing apparatus according to claim 4 , wherein
the training image selector is configured to acquire, as the training image, the partial image positioned at a random location in the area.

6. The information processing apparatus according to claim 5 , further comprising:
a first coordinate selector configured to randomly select a coordinate position in the area in a first direction; and
a second coordinate selector configured to randomly select a coordinate position in the area in a second direction intersecting the first direction, wherein
the training image selector is configured select, as the training image, the partial image having a predetermined size including the coordinate positions selected by the first coordinate selector and the second coordinate selector from the image in the area.

7. The information processing apparatus according to claim 6 , wherein
the training image selector is configured to select, as the training image, the partial image having the predetermined size with the coordinate positions selected by the first coordinate selector and the second coordinate selector as center coordinates.

8. The information processing apparatus according to claim 7 , wherein
the first coordinate selector is configured to randomly select a plurality of coordinate positions in the area in the first direction,
the second coordinate selector is configured to randomly select a plurality of coordinate positions in the area in the second direction, and
the training image selector is configured to select, as a plurality of the training images, a plurality of the partial images having a predetermined size each including the plurality of coordinate positions selected by the first coordinate selector and the second coordinate selector from the image in the area.

9. The information processing apparatus according to claim 4 , wherein
the area selector is configured to sequentially select a plurality of the areas provided at different locations in the entire area image, and
the learner is configured to update parameters of the calculation model based on differences between a plurality of label images generated by inputting a plurality of the partial images selected from the plurality of areas into the calculation model and a plurality of the corresponding reference label images.

10. The information processing apparatus according to claim 1 , wherein
the training image selector is configured to select, as the training image, the partial image selected from a random location in the entire area image.

11. The information processing apparatus according to claim 10 , further comprising:
a first coordinate selector configured to randomly select a coordinate position in the entire area image in a first direction; and
a second coordinate selector configured to randomly select a coordinate position in the entire area image in a second direction intersecting the first direction, wherein
the training image selector is configured to select, as the training image, the partial image having a predetermined size including the coordinate positions selected by the first coordinate selector and the second coordinate selector from the image in the entire area image.

12. An information processing method comprising:
acquiring an entire area image obtained by capturing an entire area of a processing surface of a wafer including at least one defect;
selecting, as a training image, a partial image including at least one defect from the entire area image;
constructing a calculation model of generating a label image obtained by extracting and binarizing the defect included in the partial image; and
updating a parameter of the calculation model based on a difference between the label image generated by inputting the training image to the calculation model and a reference label image obtained by extracting and binarizing the defect of the training image.

13. The information processing method according to claim 12 , wherein
the selecting the partial image comprises acquiring, as a plurality of the training images, a plurality of the partial images in which pixel positions of at least a part in the entire area image are different and each of which includes at least one defect, and
the updating the parameter further comprises:
sequentially calculating differences between a plurality of the label images generated by sequentially inputting the plurality of training images to the calculation model and the corresponding reference label images, and
sequentially updating parameters of the calculation model.

14. The information processing method according to claim 12 , wherein
the selecting the partial image comprises selecting, as the training image, the partial image included in an area selected from the entire area image.

15. The information processing method according to claim 14 , wherein
the selecting the partial image further comprises:
selecting the area from the entire area image, and
selecting, as the training image, the partial image included in the area selected from the entire area image.

16. The information processing method according to claim 15 , wherein
the training image selector comprises acquiring, as the training image, the partial image positioned at a random location in the area.

17. The information processing method according to claim 16 , wherein
the selecting the partial image further comprises:
randomly selecting a coordinate position in the area in a first direction;
randomly selecting a coordinate position in the area in a second dir

Assignees

• Kioxia Corp

Inventors

• Ng Youyang
• Wang Bo
• Ohashi Takuji
• Yamane Osamu
• Fujiwara Takeshi

Classifications

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T7/001 Primary

  using an image reference approach · CPC title

• G06F18/2155

  characterised by the incorporation of unlabelled data, e.g. multiple instance learning [MIL], semi-supervised techniques using expectation-maximisation [EM] or naïve labelling · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/30148

  Semiconductor; IC; Wafer · CPC title