Method and devices for indexing and seeking digital images taking into account the definition of regions of interest

• US 6,782,395 B2
• Filed: 12/01/2000
• Issued: 08/24/2004
• Est. Priority Date: 12/03/1999
• Status: Expired due to Fees

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Labelle, Lilian
• Primary Examiner(s): Rones, Charles
• Assistant Examiner(s): Mahmoudi, Hassan
• Application Number:
  US09/726,513
• Publication Number:
  US 20010003182A1
• Time in Patent Office: 1,362 Days
• Field of Search: 707/5, 707/104.1, 707/1, 707/3, 707/10, 707/100, 382/209, 358/1.14, 709/202, 709/201
• US Class Current: 1/1
• CPC Class Codes: G06F 16/583 using metadata automatically derived from the content
• CPC Class Codes: G06V 10/507 Summing image-intensity values; Histogram projection analysis
• CPC Class Codes: G06V 10/758 Involving statistics of pixels or of feature values, e.g. histogram matching
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing
• CPC Class Codes: Y10S 707/99948 Application of database or data structure, e.g. distributed, multimedia, or image

Abstract

A method of seeking digital images from among a plurality of images stored in a database by using an example image. Each of the plurality of stored images in the database is associated with a data item representing at least one characteristic of the visual content of the image.

Claims

1. A method of searching images, from an example image, from amongst a plurality of images stored in a database, each of the stored images being associated with a data item of a first type representing at least one characteristic of the visual content of the image, each of the stored images being associated with a data item of a second type representing a location of at least one region of interest in the image, said method comprising the following steps:

receiving a data item of the second type representing the location of at least one region of interest in the example image;

receiving a data item of the first type associated with the example image, the data item of the first type being obtained for each of the stored images and for the example image by performing a first operation of dividing the image into a predefined number N of blocks followed by a second extraction operation for each of the blocks (B_i) of a data item representing a visual content of the block, the data item of the first type including of a vector (G(Im)) having N components each of which corresponds to a data item representing a visual content extracted for a block of the image;

calculating a similarity between the example image and each image in at least one subset of the stored images, the similarity being calculated from the data of the first type and of the second type associated respectively with the example image and with a stored image;

said step of calculating a similarity, denoted SIM, between the example image, denoted Q, and each image, denoted D, from the least one subset of stored images being performed by calculating a distance between a vector G(Q) associated with the example image and a vector G(D) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients W_i, a value of each of which is indicative of a degree of proximity in one of the blocks B_i resulting from the division of the images, of regions of interest associated with the example image and with the stored image under consideration; and

supplying at least one image, referred to as a result image, of the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

2. An image search method according to claim 1, wherein said step of calculating a similarity SIM is effected by calculating a sum of values, each of which, denoted dist(q_i, d_i), being a measure of similarity between a component q_i of the vector G(Q) corresponding to a given block B_i of the example image and the component d_i of the vector G(D) corresponding to the same block of the stored image under consideration, the similarity measurements dist(q_i, d_i) being weighted by the weighting coefficients W_i having a value indicative of a degree of proximity, in the block, of regions of interest associated with the example image and with the stored image under consideration.

3. An image search method according to claim 2, wherein the similarity SIM is obtained by means of the following formula:

$\mathrm{SIM}=\frac{\underset{i=1}{\overset{N}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>W_i<br><br>\mathrm{dist}<br><br>\left(\mathrm{qi},\mathrm{di}\right)}{\underset{i=1}{\overset{N}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>W_i}.$

4. An image search method according to claim 1, wherein the weighting coefficients W_i have a value which is a function of a surface area of an intersection between regions of interest associated with the example image, the regions of interest associated with the stored image under consideration, and the block under consideration of the example image and of the stored image under consideration.

5. An image search method according to claim 4, in which the example image Q contains any integer number R of regions of interest, denoted ROI_r(Q), r being an integer between 1 and R, and the stored image under consideration D contains any integer number T of regions of interest, denoted ROI_t(D), t being an integer between 1 and T, in which each of the weighting coefficients W_i is obtained in accordance with the following formula:

$\begin{array}{c}{W}_i=\underset{t=1}{\overset{T}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^{r,t}\\ \mathrm{with}\\ P_i^{r,t}=\frac{\mathrm{area}<br><br>\left(B_i\bigcap <br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{ROI}}_r<br><br>\left(Q\right)\bigcap <br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{ROI}}_t<br><br>\left(D\right)\right)}{\mathrm{area}<br><br>\left(B_i\right)}\end{array}$

the index i being an integer number between 1 and the predefined number N of blocks.

6. An image search method according to claim 1, wherein the data item representing the visual content of a block of the image represents a distribution of colors (H_i) in the block.

7. An image search method according to claim 6, wherein each measurement of similarity dist(q_i, d_i) between a component d_i of the vector G(Q), corresponding to a given block B_i of the example image and the component d_i of the vector G(D) corresponding to the same block of the stored image under consideration, is obtained by calculating an intersection between these two components.

8. An image search method according to claim 1, wherein the data item of the second type representing the location of at least one region of interest in each of the stored images and in the example image includes a set of two-dimensional points indicative of a shape of the at least one region of interest and its location in an image plane of the image.

9. An image search method according to claim 1, wherein the predefined number N of blocks is equal to sixteen.

10. A method of indexing digital images, comprising the following steps, for each image Im to be indexed:

dividing an image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of the at least one region of interest;

associating with the image Im an index composed of a first vector G(Im), having N components, each of which is one of the data items of the first type, and a second vector W(Im), having N components, each of which is one of the first coefficients W_i^Im,

in which the image Im contains any integer number R of regions of interest, denoted ROI_r, r being an integer between 1 and R, and wherein in said step of calculating a first coefficient for each of the blocks B_i, the first coefficient W_i^Im is obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block B_i under consideration of the image plane of the image Im.

11. An image indexing method according to claim 10, wherein each coefficient P_i^r is obtained by calculating a ratio between a surface area of an intersection between the block B_i under consideration and the corresponding region of interest ROI_r, and the surface area of the block B_i under consideration.

12. An image indexing method according to claim 10, wherein the data item of the second type representing the location of the at least one region of interest in the image includes a set of two-dimensional points indicative of a shape of the at least one region of interest and its location in the image plane of the image.

13. An image indexing method according to claim 10, wherein the data item of the first type representing at least one characteristic of the visual content of a block of the image represents a distribution of colors in the block.

14. An image indexing method according to claim 10, wherein the image plane of the image Im is divided in accordance with a rectangular grid.

15. An image indexing method according to claim 14, wherein the predefined number N of blocks is equal to sixteen.

16. A method of searching images, from an example image, amongst a plurality of images stored in a database, wherein the example image and each of the stored images are indexed according to an image indexing method according to claim 10, comprising the following steps:

calculating a similarity SIM between the example image Q and each of the images D from amongst at least one subset of stored images, the similarity being calculated from first vectors G(Q), G(D), and second vectors W(Q), W(D), associated respectively with the example image Q and the stored image D, this step being performed by calculating a distance between the first vector G(Q) associated with the example image and the first vector G(D) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients R_i, the value of each of which is calculated as a function of the components W_i^Q, W_i^D of the second vectors W(Q) and W(D) associated respectively with the example image Q and the stored image D under consideration; and

supplying at least one image, referred to as a result image, in the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

17. An image search method according to claim 16, wherein a value of the weighting coefficients R_i is obtained in accordance with the following formulae:

R_i=W_i^Q+W_i^D if W_i^Q and W_i^D are both strictly greater than the predefined constant W₀ and

R_i=W_i^Q+W_i^D−

W₀ in other cases.

18. An image search method according to claim 16, wherein the similarity SIM is obtained by calculating a sum of values, each of which, denoted dist(q_i, d_i), is a measurement of similarity between a component q_i of the first vector G(Q) corresponding to a given block B_i of the example image Q and the component d_i of the first vector G(D) corresponding to the same block of the stored image D under consideration, these similarity measurements dist(q_i, d_i) being weighted by the weighting coefficients R_i.

19. An image search method according to claim 18, wherein the similarity SIM is obtained by means of the following formula:

$\mathrm{SIM}=\frac{\underset{i=1}{\overset{N}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>R_i<br><br>\mathrm{dist}<br><br>\left(\mathrm{qi},\mathrm{di}\right)}{\underset{i=1}{\overset{N}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>R_i}.$

20. An image search method according to claim 19, wherein each of the measurements of similarity, dist(q_i, d_i), between a component q_i of the vector G(Q) corresponding to a given block B_i of the example image Q and the component d_i of the vector G(D) corresponding to the same block in the stored image under consideration D, is obtained by calculating an intersection between these two components.

21. A device for searching for images, from an example image, from amongst a plurality of images stored in a database, said device comprising means adapted to implement each step of an image search method, said method being for searching images from an example image, from amongst a plurality of images stored in a database, each of the stored images being associated with a data item of a first type representing at least one characteristic of the visual content of the image, each of the stored images being associated with a data item of a second type representing a location of at least one region of interest in the image, said method comprising the following steps:

receiving a data item of the second type representing the location of at least one region of interest in the example image;

receiving a data item of the first type associated with the example image, the data item of the first type being obtained for each of the stored images and for the example image by performing a first operation of dividing the image into a predefined number N of blocks followed by a second extraction operation for each of the blocks (B_i) of a data item representing a visual content of the block, the data item of the first type including of a vector (G(Im)) having N components each of which corresponds to a data item representing a visual content extracted for a block of the image;

calculating a similarity between the example image and each image in at least one subset of the stored images, the similarity being calculated from the data of the first type and of the second type associated respectively with the example image and with a stored image,

said step of calculating a similarity, denoted SIM, between the example image, denoted Q, and each image, denoted D, from the least one subset of stored images being performed by calculating a distance between a vector G(Q) associated with the example image and a vector G(D) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients W_i, a value of each of which is indicative of a degree of proximity in one of the blocks B_i resulting from the division of the images, of regions of interest associated with the example image and with the stored image under consideration; and

supplying at least one image, referred to as a result image, of the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

22. A device for searching images, from an example image, amongst a plurality of images stored in a database, said device comprising means adapted to implement each step of an image indexing method for indexing digital images, comprising the following steps, for each image Im to be indexed:

dividing an image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of the at least one region of interest;

associating with the image Im an index composed of a first vector G(Im), having N components, each of which is one of the data items of the first type, and a second vector W(Im), having N components, each of which is one of the first coefficients W_i^Im,

in which the image Im contains any integer number R of regions of interest, denoted ROI_r, r being an integer between 1 and R, and wherein in said step of calculating a first coefficient for each of the blocks B_i, the first coefficient W_i^Im is obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block B_i under consideration of the image plane of the image Im.

23. A device for searching images, from an example image, amongst a plurality of images stored in a database, the images being indexed according to an indexing method comprising the following steps, for each image Im to be indexed:

dividing an image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of the at least one region of interest;

associating with the image Im an index composed of a first vector G(Im), having N components, each of which is one of the data items of the first type, and a second vector W(Im), having N components, each of which is one of the first coefficients W_i^Im,

in which the image Im contains any integer number R of regions of interest, denoted ROI_r, r being an integer between 1 and R, and wherein in said step of calculating a first coefficient for each of the blocks B_i, the first coefficient W_i^Im is obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block B_i under consideration of the image plane of the image Im,

said device comprising means adapted to implement each step of an image search method for searching images from the example image, amongst a plurality of images stored in a database, wherein the image searching method comprises the following steps;

calculating a similarity SIM between the example image Q and each of the images D from amongst at least one subset of stored images, the similarity being calculated from first vectors G(Q), G(D), and second vectors W(Q), W(D), associated respectively with the example image Q and the stored image D, this step being performed by calculating a distance between the first vector G(Q) associated with the example image and the first vector G(Q) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients R_i, the value of each of which is calculated as a function of the components W_i^Q, W_i^D of the second vectors W(Q) and W(D) associated respectively with the example image Q and the stored image D under consideration; and

supplying at least one image, referred to as a result image, in the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

24. A device for searching images, from an example image, from amongst a plurality of images stored in a database, each of the stored images being associated with a data item of a first type representing at least one characteristic of the visual content of the image, each of the stored images being associated with a data item of a second type representing the location of at least one region of interest in the image, said device comprising:

means for receiving a data item of the second type representing the location of at least one region of interest in the example image;

means for receiving a data item of the first type associated with the example image, the data item of the first type being obtained for each of the stored images and for the example image by performing a first operation of dividing the image into a predefined number N of blocks followed by a second extraction operation for each of the blocks (B_i) of a data item representing a visual content of the block, the data item of the first type including of a vector (G(Im)) having N components each of which corresponds to a data item representing a visual content extracted for a block of the image;

means for calculating a similarity between the example image and each of the images amongst at least one subset of the stored images, the similarity being calculated from the data of the first type and of the second type associated respectively with the example image and with a stored image,

said means for calculating a similarity between the example image Q and each of the images D from amongst at least one subset of stored images including means for calculating a distance between the vector G(Q) associated with the example image and the vector G(D) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients W_i, a value of each of which being indicative of a degree of proximity in one of the blocks B_i resulting from the division of the images, of regions of interest associated with the example image and with the stored image under consideration; and

means of supplying at least one image, referred to as a result image, of the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

25. A device for indexing digital images, where an image Im to be indexed contains any integer number R of regions of interest, denoted ROI_r, r being an integer between 1 and R, said device comprising:

means for dividing the image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

means for extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

means for defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

means for calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of the at least one region of interest, the first coefficient W_i^Im being obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block G_i under consideration of the image plane of the image Im; and

means for associating with the image Im an index composed of a first vector G(Im), having N components, each of which being one of the data items of the first type, and a second vector W(Im), having N components, each of which being one of the first coefficients W_i^Im.

26. A computer, comprising means adapted to implement each step of an image search method, said method being for searching images from an example image, from amongst a plurality of images stored in a database, each of the stored images being associated with a data item of a first type representing at least one characteristic of the visual content of the image, each of the stored images being associated with a data item of a second type representing a location of at least one region of interest in the image, said method comprising the following steps:

receiving a data item of the second type representing the location of at least one region of interest in the example image;

receiving a data item of the first type associated with the example image, the data item of the first type being obtained for each of the stored images and for the example image by performing a first operation of dividing the image into a predefined number N of blocks followed by a second extraction operation for each of the blocks (B_i) of a data item representing a visual content of the block, the data item of the first type including of a vector (G(Im)) having N components each of which corresponds to a data item representing a visual content extracted for a block of the image;

calculating a similarity between the example image and each image in at least one subset of the stored images, the similarity being calculated from the data of the first type and of the second type associated respectively with the example image and with a stored image,

said step of calculating a similarity, denoted SIM, between the example image, denoted Q, and each image, denoted D, from the least one subset of stored images being performed by calculating a distance between a vector G(Q) associated with the example image and a vector G(Q) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients W_i, a value of each of which is indicative of a degree of proximity in one of the blocks B_i resulting from the division of the images, of regions of interest associated with the example image and with the stored image under consideration; and

supplying at least one image, referred to as a result image, of the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

27. A computer, comprising an image search device for searching for images, from an example image, from amongst a plurality of images stored in a database, said device comprising means adapted to implement each step of an image search method, said method being for searching images from an example image, from amongst a plurality of images stored in a database, each of the stored images being associated with a data item of a first type representing at least one characteristic of the visual content of the image, each of the stored images being associated with a data item of a second type representing a location of at least one region of interest in the image, said method comprising the following steps:

receiving a data item of the second type representing the location of at least one region of interest in the example image;

receiving a data item of the first type associated with the example image, the data item of the first type being obtained for each of the stored images and for the example image by performing a first operation of dividing the image into a predefined number N of blocks followed by a second extraction operation for each of the blocks (B_i) of a data item representing a visual content of the block, the data item of the first type including of a vector (G(Im)) having N components each of which corresponds to a data item representing a visual content extracted for a block of the image;

calculating a similarity between the example image and each image in at least one subset of the stored images, the similarity being calculated from the data of the first type and of the second type associated respectively with the example image and with a stored image,

said step of calculating a similarity, denoted SIM, between the example image, denoted Q, and each image, denoted D, from the least one subset of stored images being performed by calculating a distance between a vector G(Q) associated with the example image and a vector G(D) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients W_i, a value of each of which is indicative of a degree of proximity in one of the blocks B_i resulting from the division of the images, of regions of interest associated with the example image and with the stored image under consideration; and

supplying at least one image, referred to as a result image, of the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

28. A computer, comprising means adapted to implement each step of an image indexing method for indexing digital images, comprising the following steps, for each image Im to be indexed:

dividing an image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of the at least one region of interest;

associating with the image Im an index composed of a first vector G(Im), having N components, each of which is one of the data items of the first type, and a second vector W(Im), having N components, each of which is one of the first coefficients W_i^Im,

in which the image Im contains any integer number R of regions of interest, denoted ROI_r, being an integer between 1 and R, and wherein in said step of calculating a first coefficient for each of the blocks B_i, the first coefficient W_i^Im is obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block B_i under consideration of the image plane of the image Im.

29. A computer, comprising an image search device for searching images from an example image, amongst a plurality of images stored in a database, each of the images having been indexed according to an image indexing method for indexing digital images, comprising the following steps, for each image Im to be indexed:

dividing an image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of the at least one region of interest;

associating with the image Im an index composed of a first vector G(Im), having N components, each of which is one of the data items of the first type, and a second vector W(Im), having N components, each of which is one of the first coefficients W_i^Im,

in which the image Im contains any integer number R of regions of interest, denoted ROI_r, r being an integer between 1 and R, and wherein in said step of calculating a first coefficient for each of the blocks B_i, the first coefficient W_i^Im is obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block B_i under consideration of the image plane of the image Im, and

wherein said device comprises means for implementing each step of an image search method for searching images from the example image, amongst a plurality of images stored in a database, wherein the image search method comprises the following steps;

calculating a similarity SIM between the example image Q and each of the images D from amongst at least one subset of stored images, the similarity being calculated from first vectors G(O), G(D), and second vectors W(Q), W(D), associated respectively with the example image Q and the stored image D, this step being performed by calculating a distance between the first vector G(Q) associated with the example image and the first vector G(D) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients R_i, the value of each of which is calculated as a function of the components W_i^Q, W_i^D of the second vectors W(Q) and W(D) associated respectively with the example image Q and the stored image D under consideration; and

supplying at least one image, referred to as a result image, in the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

30. A computer, comprising means adapted to implement an image search method, each of the images being searched having been indexed according to an indexing method comprising the following steps, for each image Im to be indexed:

dividing an image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of the at least one region of interest;

associating with the image Im an index composed of a first vector G(Im), having N components, each of which is one of the data items of the first type, and a second vector W(Im), having N components, each of which is one of the first coefficients W_i^Im,

in which the image Im contains any integer number R of regions of interest, denoted ROI_r, r being an integer between 1 and R, and wherein in said step of calculating a first coefficient for each of the blocks B_i, the first coefficient W_i^Im is obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block B_i under consideration of the image plane of the image Im.

31. A computer program which can be loaded into programmable apparatus, comprising sequences of instructions or portions of software code for implementing the following steps for searching images, from an example image, from amongst a plurality of images stored in a database, each of the stored images being associated with a data item of a first type representing at least one characteristic of the visual content of the image, each of the stored images being associated with a data item of a second type representing the location of at least one region of interest in the image, when loaded into and executed from by the programmable apparatus:

receiving a data item of the second type representing the location of at least one region of interest in the example image;

receiving a data item of said first type associated with the example image, the data item of the first type being obtained for each of the stored images and for the example image by performing a first operation of dividing the image into a predefined number N of blocks followed by a second extraction operation for each of the blocks (B_i) of a data item representing the visual content of the block, the data item of the first type consisting of a vector (G(Im)) having N components each of which being the data item representing the visual content extracted for a block of the image;

calculating a similarity between the example image and each of the images amongst at least one subset of the stored images, the similarity being calculated from the data of the first type and of the second type associated respectively with the example image and with a stored image,

said step of calculating a similarity, denoted SIM, between the example image Q and each of the images D from amongst at least one subset of stored images being performed by calculating a distance between the vector G(Q) associated with the example image and the vector G(D) associated with the stored image under consideration, the distance being weighted by means of a plurality N of weighting coefficients W_i, the value of each of which being indicative of a degree of proximity in one of the blocks B_i resulting from the division of the images, of regions of interest associated with the example image and with the stored image under consideration; and

supplying at least one image, referred to as a result image, of the database, the at least one result image being selected from amongst the stored images in the database according to its degree of similarity with the example image.

32. A computer program which can be loaded into a programmable apparatus, comprising sequences of instructions or portions of software code for implementing the following steps for indexing digital images, when loaded into and executed by the programmable apparatus, where an image Im to be indexed contains any integer number R of regions of interest, denoted ROI_r, r being an integer between 1 and R:

dividing the image plane of the image Im, in accordance with a partitioning comprising a predefined number N of blocks B_i;

extracting from each of the blocks a data item of a first type representing at least one characteristic of a visual content of the block;

defining in the image at least one region of interest, and obtaining a data item of a second type representing a location of at least one region of interest in the image;

calculating for each of the blocks B_i a first coefficient, W_i^Im, indicating a degree of presence in the block B_i under consideration of the image of said at least one region of interest, the first coefficient W_i^Im being obtained by means of the following formula;

$W_i^{I<br><br>\text{\hspace{1em}<br><br>}<br><br>m}=W_o+\underset{r=1}{\overset{R}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>P_i^r$

in which W₀ is a predefined constant, and P_i^r is a coefficient indicating a degree of presence of a corresponding region of interest ROI_r, in the block G_i under consideration of the image plane of the image Im; and

associating with the image Im an index composed of a first vector G(Im), having N components, each of which being one of the data items of the first type, and a second vector W(Im), having N components, each of which being one of the first coefficients W_i^Im.