Image signature extraction device

Patent metadata

Field	Value
Publication number	US-10133956-B2
Application number	US-201013254775-A
Country	US
Kind code	B2
Filing date	Mar 12, 2010
Priority date	Mar 13, 2009
Publication date	Nov 20, 2018
Grant date	Nov 20, 2018

Abstract

The image signature extraction device includes an image signature generation unit and an encoding unit. The image signature generation unit extracts region features from respective sub-regions in an image in accordance with a plurality of pairs of sub-regions in the image, the pairs of sub-regions including at least one pair of sub-regions in which both a combination of shapes of two sub-regions of the pair and a relative position between the two sub-regions of the pair differ from those of at least one of other pairs of sub-regions, and based on the extracted region features of the respective sub-regions, generates an image signature to be used for identifying the image. The encoding unit encodes the image signature.

First claim

The invention claimed is:

1. An image signature extraction device, comprising:
a memory configured to store a program and at least one processor configured to read the program to execute:
a calculation unit configured to calculate region features from a plurality of pairs of sub-regions in an image, the pairs of sub-regions being associated with each dimension composing an image signature, the image signature being information for identifying an image;
a quantization unit configured to quantize difference values between the region features into one of three values, as a value of each dimension of the image signature, using a threshold determined based on absolute values of the difference values; and
an encoding unit configured to map 243 different combinations of quantization values of five dimensions to values from 0 to 242 and to encode quantization values of five dimensions into 8 bit,
wherein the pairs of sub-regions including at least one pair of sub-regions in which both a combination of shapes of the two sub-regions of the pair and a relative position between the two sub-regions of the pair differ from those of at least one of other pairs of sub-regions; and
wherein the image signature identifying the image and a second image signature identifying a second image are matched by calculating an identity scale indicating a distance or a similarity value that the image and the second image are identical.

2. The image signature extraction device, according to claim 1 , wherein the quantization unit quantizes the difference values into the one of three values that include a first quantization value, a second quantization value, which is a value smaller than the first quantization value, and a third quantization value, which is a value smaller than the second quantization value, and
a difference between the first quantization value and the second quantization value and a difference between the second quantization value and the third quantization value are equal.

3. A matching device that performs matching using the image signature output by the image signature extraction device according to claim 1 .

4. An identifying device that performs identification using the image signature output by the image signature extraction device according to claim 1 .

5. The image signature extraction device, according to claim 1 , wherein the image is a frame of a video.

6. An image signature extraction method performed by at least one processor configured to read a program stored on a memory to perform the method, the method comprising:
calculating region features from a plurality of pairs of sub-regions in an image, the pairs of sub-regions being associated with each dimension composing an image signature, the image signature being information for identifying an image;
quantizing the difference values between the region features into one of three values, as a value of each dimension of the image signature, using a threshold determined based on absolute values of the difference values;
mapping 243 different combinations of quantization values of five dimensions to values from 0 to 242; and
encoding quantization values of five dimensions into 8 bit,
wherein the pairs of sub-regions including at least one pair of sub-regions in which both a combination of shapes of the two sub-regions of the pair and a relative position between the two sub-regions of the pair differ from those of at least one of other pairs of sub-regions; and
wherein the image signature identifying the image and a second image signature identifying a second image are matched by calculating an identity scale indicating a distance or a similarity value that the image and the second image are identical.

7. The image signature extraction method, according to claim 6 , wherein the quantizing includes quantizing the difference values into the one of three values that include a first quantization value, a second quantization value, which is a value smaller than the first quantization value, and a third quantization value, which is a value smaller than the second quantization value, and
a difference between the first quantization value and the second quantization value and a difference between the second quantization value and the third quantization value are equal.

8. A matching method for performing matching using the image signature extracted by the image signature extraction method according to claim 6 .

9. An identifying method for performing identification using the image extracted by the image signature extraction method according to claim 6 .

10. A non-transitory computer-readable medium storing a program comprising instructions for causing a computer to function as:
a calculation unit that calculates region features from a plurality of pairs of sub-regions in an image, the pairs of sub-regions being associated with each dimension composing an image signature, the image signature being information for identifying an image;
a quantization unit that quantizes the difference values between the region features into one of three values, as a value of each dimension of the image signature, using a threshold determined based on absolute values of the difference values; and
an encoding unit that maps 243 different combinations of quantization values of five dimensions to values from 0 to 242 and encodes quantization values of five dimensions into 8 bit,
wherein the pairs of sub-regions including at least one pair of sub-regions in which both a combination of shapes of the two sub-regions of the pair and a relative position between the two sub-regions of the pair differ from those of at least one of other pairs of sub-regions; and
wherein the image signature identifying the image and a second image signature identifying a second image are matched by calculating an identity scale indicating a distance or a similarity value that the image and the second image are identical.

11. A matching device comprising:
at least one memory configured to store at least one program and a lookup table;
an image feature extraction device; and
at least one processor configured to read the at least one program to cause the image feature extraction device to execute:
a calculation unit configured to calculate region features from a plurality of pairs of sub-regions in an image, the pairs of sub-regions being associated with each dimension composing an image signature, the image signature being information for identifying an image;
a quantization unit configured to quantize difference values between the region features into one of three values, as a value of each dimension of the image signature, using a threshold determined based on absolute values of the difference values; and
an encoding unit configured to map 243 different combinations of quantization values of five dimensions to values from 0 to 242 and to encode quantization values of five dimensions into 8 bit,
wherein the pairs of sub-regions including at least one pair of sub-regions in which both a combination of shapes of the two sub-regions of the pair and a relative position between the two sub-regions of the pair differ from those of at least one of other pairs of sub-regions,
wherein the image signature identifying the image and a second image signature identifying a second image are matched by calculating an identity scale indicating a distance or a similarity value that the image and the second image are identical, and
wherein the distance or the similarity value between a value of the image and a value of the second image is acquired from a lookup table in which the distance or the similarity value between the values of the image and the second image is calculated and stored.

Assignees

• Iwamoto Kota
• Oami Ryoma
• Nec Corp

Inventors

• Iwamoto Kota
• Oami Ryoma

Classifications

• G06V10/50 Primary

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title

• G06V10/421

  by analysing segments intersecting the pattern · CPC title

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06V10/56

  relating to colour · CPC title

• G06F16/5854

  using shape and object relationship · CPC title