Back to results
US 9,754,191 B2Granted
Feature image generation apparatus, classification apparatus and non-transitory computer-readable memory, and feature image generation method and classification method
AssigneeMegaChips Corporation
InventorsKenta Nishiyuki, Kenta Nagamine
Life of this patent
Estimating the term…
Abstract
A feature image generation apparatus includes circuitry. The circuitry generates, on the basis of a processing target image in which an object appears, a first image showing the object, and generates, as a feature image showing a feature of the object, at least a part of a rotational composite image obtained by composition of a plurality of rotated images obtained by rotating the first image.
Claims5
A classification apparatus comprising:
- circuitry configured to:
- generate, based on a processing target image, a first image showing an object, the processing target image being generated by an image capturing circuit that captures an image of the object whose circumferential orientation is not constant when the image capturing circuit captures the image;
- generate, as a feature image showing a feature of the object, at least a part of a piece of a first rotational combined image obtained by combining a plurality of first rotated images obtained by rotating the first image;
- compare the feature image with a template feature image showing a feature of a comparison target and classify, based on a comparison result, whether the object appearing in the processing target image corresponds to the comparison target, wherein the template feature image is a composite image obtained by composition of a plurality of standard feature images individually generated based on a plurality of standard images in which a plurality of comparison targets individually appear, and each of the plurality of standard feature images is at least a part of a second rotational combined image obtained by combining a plurality of second rotated images obtained by rotating a second image that is generated based on the standard image used in the generation of the standard feature image and shows a comparison target appearing in the standard image;
- and
- classify, based on the feature image, whether the object appearing in the processing target image is proper, the object being one of a coin, a substrate on which a plurality of components are mounted, a snack, and a sticker attached to a package.
3 more elements
A method for classification implemented by circuitry, the method comprising:
- generating, based on a processing target image, a first image showing an object, the processing target image being generated by an image capturing circuit that captures an image of the object whose circumferential orientation is not constant when the image capturing circuit captures the image;
- generating, as a feature image showing a feature of the object, at least a part of a piece of a first rotational combined image obtained by combining a plurality of first rotated images obtained by rotating the first image;
- comparing the feature image with a template feature image showing a feature of a comparison target and classify, based on a comparison result, whether the object appearing in the processing target image corresponds to the comparison target, wherein the template feature image is a composite image obtained by composition of a plurality of standard feature images individually generated based on a plurality of standard images in which a plurality of comparison targets individually appear, and each of the plurality of standard feature images is at least a part of a second rotational combined image obtained by combining a plurality of second rotated images obtained by rotating a second image that is generated based on the standard image used in the generation of the standard feature image and shows a comparison target appearing in the standard image;
- and
- classifying, based on the feature image, whether the object appearing in the processing target image is proper, the object being one of a coin, a substrate on which a plurality of components are mounted, a snack, and a sticker attached to a package.
2 more elements
Dependencies are inferred from the claim text.