Image retrieval apparatus, control method for the same, and storage medium
AssigneeCanon Kabushiki Kaisha
InventorsKoichi Magai, Hirotaka Shiiyama
Life of this patent
Estimating the term…
Abstract
An image retrieval apparatus configured so as to enable a global feature method and a local feature method to complement each other is provided. After obtaining a retrieval result candidate using the local feature method, the image retrieval apparatus further verifies global features already registered in a database, with regard to the retrieval result candidate image. A verification position of the global features is estimated using the local features.
Claims20
An image retrieval apparatus, comprising:
- an extraction unit configured to extract a local feature amount, which is a localized feature amount, from an input image;
- a retrieval unit configured to calculate a local feature similarity, which is a similarity between the extracted local feature amount and a local feature amount of each image registered in an image feature database, and acquire a plurality of candidate images based on the calculated local feature similarity;
- a parameter calculation unit configured to calculate a geometric transformation parameter between each candidate image and the input image using the extracted local feature amount;
- a determination unit configured to determine a calculation region on the input image corresponding to a global feature amount calculation region calculated in advance with respect to each candidate image, using the geometric transformation parameter;
- a feature amount calculation unit configured to calculate a global feature amount in the calculation region determined by the determination unit;
- a similarity calculation unit configured to calculate a similarity between each candidate image and the input image by comparing the global feature amount of each candidate image and the global feature amount of the input image;
- and
- an output unit configured to output each candidate image in order of similarity.
A control method for an image retrieval apparatus, comprising:
- an extraction step of extracting a local feature amount, which is a localized feature amount, from an input image;
- a retrieval step of calculating a local feature similarity, which is a similarity between the extracted local feature amount and a local feature amount of each image registered in an image feature database, and acquiring a plurality of candidate images based on the calculated local feature similarity;
- a parameter calculation step of calculating a geometric transformation parameter between each candidate image and the input image using the extracted local feature amount;
- a determination step of determining a calculation region on the input image corresponding to a global feature amount calculation region calculated in advance with respect to each candidate image, using the geometric transformation parameter;
- a feature amount calculation step of calculating a global feature amount in the calculation region determined by the determination unit;
- a similarity calculation step of calculating a similarity between each candidate image and the input image by comparing the global feature amount of each candidate image and the global feature amount of the input image;
- and
- an output step of outputting each candidate image in order of similarity.
An image retrieval apparatus, comprising:
- a local feature extraction unit configured to extract a local feature amount, which is a localized feature amount, from an input image;
- an acquisition unit configured to calculate a local feature similarity, which is a similarity between the extracted local feature amount, and a local feature amount of each image registered in an image feature database, and acquire a plurality of candidate images based on the calculated local feature similarity;
- an additional feature extraction unit configured to extract an additional feature, which is an added feature, from each of the input image and the acquired plurality of candidate images;
- a calculation unit configured to calculate an additional feature similarity, which is a similarity between the additional feature extracted from each of the plurality of candidate images, and the additional feature extracted from the input image;
- and
- an output unit configured to calculate a general similarity using the local feature similarity and the additional feature similarity, and output a predetermined number of candidate images as a retrieval result in order of calculated general similarity.
A control method for an image retrieval apparatus, comprising:
- a local feature extraction step of extracting a local feature amount, which is a localized feature amount, from an input image;
- an acquisition step of calculating a local feature similarity, which is a similarity between the extracted local feature amount, and a local feature amount of each image registered in an image feature database, and acquiring a plurality of candidate images based on the calculated local feature similarity;
- an additional feature extraction step of extracting an additional feature, which is an added feature, from each of the input image and the acquired plurality of candidate images;
- a calculation step of calculating an additional feature similarity, which is a similarity between the additional feature extracted from each of the plurality of candidate images, and the additional feature extracted from the input image;
- and
- an output step of calculating a general similarity using the local feature similarity and the additional feature similarity, and outputting a predetermined number of candidate images as a retrieval result in order of calculated general similarity.
Dependencies are inferred from the claim text.