IPC분류정보
국가/구분 |
United States(US) Patent
등록
|
국제특허분류(IPC7판) |
|
출원번호 |
US-0690381
(2012-11-30)
|
등록번호 |
US-9183460
(2015-11-10)
|
발명자
/ 주소 |
- Zhang, John R.
- Rosenberg, Charles J.
- Song, Yang
|
출원인 / 주소 |
|
대리인 / 주소 |
|
인용정보 |
피인용 횟수 :
2 인용 특허 :
3 |
초록
▼
Methods, systems and apparatus for identifying modified images based on visual dissimilarity to a first image. In an aspect, a method includes determining, for each of a first image and a second image, a respective set of local image feature descriptions; determining one or more unmatched regions of
Methods, systems and apparatus for identifying modified images based on visual dissimilarity to a first image. In an aspect, a method includes determining, for each of a first image and a second image, a respective set of local image feature descriptions; determining one or more unmatched regions of the images that include unmatched image features and that correspond to one or more same respective regions in both the first image and the second image; determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image; and determining that the second image is a modification of the first image when one of the modification measures meets a modification measure threshold.
대표청구항
▼
1. A computer-implemented method performed by data processing apparatus, the method comprising: determining, for each of a first image and a second image, each of the first and second image comprising respective image data depicting visual content, a respective set of local image feature description
1. A computer-implemented method performed by data processing apparatus, the method comprising: determining, for each of a first image and a second image, each of the first and second image comprising respective image data depicting visual content, a respective set of local image feature descriptions, each local image feature description describing a local image feature detected in the image and the location of the local image feature within the image;determining one or more unmatched regions of the images that include unmatched image features and that correspond to one or more same respective regions in both the first image and the second image;determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image; anddetermining that the second image is a modification of the first image when one of the modification measures meets a modification measure threshold;wherein:determining one or more unmatched regions of the images comprises:determining, from the respective sets of local image feature descriptions, unmatched local image features, each unmatched local image feature being a local image feature in one of the respective sets that does not have a corresponding matching local image feature in the other set of the respective sets; anddetermining, from the unmatched local image features, one or more unmatched regions of the images that include unmatched image features and that correspond to one or more same respective regions in both the first image and the second image. 2. The computer-implemented method of claim 1, wherein: the second image is an image that has been determined to satisfy a threshold match of visual similarity to the first image. 3. The computer-implemented method of claim 2, wherein determining unmatched local image features comprises: identifying candidate unmatched local image features, each candidate unmatched local image feature being a local image feature in one of the respective sets that does not have a corresponding matching local image feature in the other set of the respective sets;for each candidate unmatched local image feature, determining a ratio based candidate unmatched local image features and matched local image features that within a sub-region that includes the candidate unmatched local image feature; andidentifying each candidate unmatched local image feature having a ratio that meets a ratio threshold as an unmatched local image feature. 4. The computer-implemented method of claim 3, wherein determining one or more unmatched regions of the images comprises: for each image: generating one or more feature clusters, each feature cluster being generated by grouping into the feature cluster unmatched local image features that are each within a respective distance of an unmatched local image feature that belongs to the feature cluster; anddetermining, for each feature clusters, a boundary that includes the unmatched local image features that belong to the feature cluster, wherein the boundary defines an unmatched region. 5. The computer-implemented method of claim 4, further comprising: generating an image mask from the unmatched regions, the image masking only data outside of each of the unmatched regions; andwherein determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image comprises: applying the image mask to each of the first image and the second image; anddetermining, for each of the one or more unmatched regions of the images, the modification measure comprises determining the modification measure for each unmatched region from only the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image. 6. The computer-implemented method of claim 5, wherein generating an image mask from the unmatched regions comprises: for each unmatched region, determining whether the region meets a minimum size threshold; andgenerating the image mask from only unmatched regions that meet the minimum size threshold. 7. The computer-implemented method of claim 5, wherein generating an image mask from the unmatched regions comprises: for each unmatched region, determining respective color histogram data for each of the first image and the second image from the respective image data corresponding to the unmatched region; andgenerating the image mask from only unmatched regions for which a correlation of the respective color histogram data does not meet a correlation threshold. 8. The computer-implemented method of claim 2, wherein determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image comprises: determining a modification measure based on first facial features detected in the image data corresponding to the unmatched region in the first image and second facial features detected in the image data corresponding to the unmatched region in the second image. 9. The computer-implemented method of claim 1, wherein determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image comprises: determining a modification measure based on first text features detected in the image data corresponding to the unmatched region in the first image and second text features detected in the image data corresponding to the unmatched region in the second image. 10. The computer-implemented method of claim 1, wherein the local image feature descriptions comprise one or more of edge descriptors and corner descriptors. 11. A system, comprising: a data processing apparatus; anda non-transitory memory storage apparatus in data communication with the data processing apparatus and storing instructions executable by the data processing apparatus and that upon such execution cause the data processing apparatus to perform operations comprising:determining, for each of a first image and a second image, each of the first and second image comprising respective image data depicting visual content, a respective set of local image feature descriptions, each local image feature description describing a local image feature detected in the image and the location of the local image feature within the image;determining one or more unmatched regions of the images that include unmatched image features and that correspond to one or more same respective regions in both the first image and the second image;determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image; anddetermining that the second image is a modification of the first image when one of the modification measures meets a modification measure threshold;wherein:determining one or more unmatched regions of the images comprises:determining, from the respective sets of local image feature descriptions, unmatched local image features, each unmatched local image feature being a local image feature in one of the respective sets that does not have a corresponding matching local image feature in the other set of the respective sets; anddetermining, from the unmatched local image features, one or more unmatched regions of the images that include unmatched local image features and that correspond to one or more same respective regions in both the first image and the second image. 12. The system of claim 11, wherein: the second image is an image that has been determined to satisfy a threshold match of visual similarity to the first image. 13. The system of claim 12, wherein determining unmatched local image features comprises: identifying candidate unmatched local image features, each candidate unmatched local image feature being a local image feature in one of the respective sets that does not have a corresponding matching local image feature in the other set of the respective sets;for each candidate unmatched local image feature, determining a ratio based candidate unmatched local image features and matched local image features that within a sub-region that includes the candidate unmatched local image feature; andidentifying each candidate unmatched local image feature having a ratio that meets a ratio threshold as an unmatched local image feature. 14. The system of claim 13, wherein determining one or more unmatched regions of the images comprises: for each image: generating one or more feature clusters, each feature cluster being generated by grouping into the feature cluster unmatched local image features that are each within a respective distance of an unmatched local image feature that belongs to the feature cluster; anddetermining, for each feature clusters, a boundary that includes the unmatched local image features that belong to the feature cluster, wherein the boundary defines an unmatched region. 15. The system of claim 14, wherein the operations further comprising: generating an image mask from the unmatched regions, the image masking only data outside of each of the unmatched regions; andwherein determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image comprises: applying the image mask to each of the first image and the second image; anddetermining, for each of the one or more unmatched regions of the images, the modification measure comprises determining the modification measure for each unmatched region from only the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image. 16. The system of claim 15, wherein generating an image mask from the unmatched regions comprises: for each unmatched region, determining whether the region meets a minimum size threshold; andgenerating the image mask from only unmatched regions that meet the minimum size threshold. 17. The system of claim 15, wherein generating an image mask from the unmatched regions comprises: for each unmatched region, determining respective color histogram data for each of the first image and the second image from the respective image data corresponding to the unmatched region; andgenerating the image mask from only unmatched regions for which a correlation of the respective color histogram data does not meet a correlation threshold. 18. A non-transitory memory storage apparatus storing instructions executable by a data processing apparatus and that upon such execution cause the data processing apparatus to perform operations comprising: determining, for each of a first image and a second image, each of the first and second image comprising respective image data depicting visual content, a respective set of local image feature descriptions, each local image feature description describing a local image feature detected in the image and the location of the local image feature within the image;determining one or more unmatched regions of the images that include unmatched image features and that correspond to one or more same respective regions in both the first image and the second image;determining, for each of the one or more unmatched regions of the images, a modification measure based on the image data corresponding to the unmatched region in the first image and the image data corresponding to the unmatched region in the second image; anddetermining that the second image is a modification of the first image when one of the modification measures meets a modification measure threshold;wherein:determining one or more unmatched regions of the images comprises:determining, from the respective sets of local image feature descriptions, unmatched local image features, each unmatched local image feature being a local image feature in one of the respective sets that does not have a corresponding matching local image feature in the other set of the respective sets; anddetermining, from the unmatched local image features, one or more unmatched regions of the images that include unmatched image features and that correspond to one or more same respective regions in both the first image and the second image.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.