GB2388761A - Texture-based retrieval of similar images - Google Patents

Texture-based retrieval of similar images Download PDF

Info

Publication number
GB2388761A
GB2388761A GB0319202A GB0319202A GB2388761A GB 2388761 A GB2388761 A GB 2388761A GB 0319202 A GB0319202 A GB 0319202A GB 0319202 A GB0319202 A GB 0319202A GB 2388761 A GB2388761 A GB 2388761A
Authority
GB
United Kingdom
Prior art keywords
texture
image
query
descriptor
images
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
GB0319202A
Other versions
GB2388761B (en
GB0319202D0 (en
Inventor
Jin Woong Kim
Ki Won You
Munchurl Kim
Yong Man Ro
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Electronics and Telecommunications Research Institute ETRI
Original Assignee
Electronics and Telecommunications Research Institute ETRI
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from KR1019990041574A external-priority patent/KR100308456B1/en
Application filed by Electronics and Telecommunications Research Institute ETRI filed Critical Electronics and Telecommunications Research Institute ETRI
Publication of GB0319202D0 publication Critical patent/GB0319202D0/en
Publication of GB2388761A publication Critical patent/GB2388761A/en
Application granted granted Critical
Publication of GB2388761B publication Critical patent/GB2388761B/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/40—Analysis of texture
    • G06T7/41—Analysis of texture based on statistical description of texture
    • G06T7/42—Analysis of texture based on statistical description of texture using transform domain methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G06F16/5862—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using texture
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/42—Global feature extraction by analysis of the whole pattern, e.g. using frequency domain transformations or autocorrelation
    • G06V10/431—Frequency domain transformation; Autocorrelation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • G06V10/7515—Shifting the patterns to accommodate for positional errors

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Medical Informatics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Library & Information Science (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Processing Or Creating Images (AREA)
  • Image Analysis (AREA)

Abstract

Retrieving images where channels obtained by transforming the stored and query images to the frequency domain and partitioning them are compared to retrieve similar images. Specifically, the distance between the texture descriptors obtained from the channels is used to measure the similarity of the images. Also disclosed is retrieving images where texture descriptors and rotation information obtained from the stored and query images is used to retrieve images. Specifically, the rotation information obtained from the query image is compared with the rotation information of the stored image, these images are then aligned using this information. The similarity of the images is then measured by comparing the distance between the texture descriptors in both images.

Description

TEXTURE DESCRIPTION METHOD Al TEXT-BASED RETRIEV:
METHOD IN FREQUENCY DOMAIN
TECHNICAL FIELD
The present invention relates to a texture description method of an
5 image, and more particularly, to a method of describing image texture in the frequency A a, n r- h, h À m a,, A _ I a A 1. + r ?, 0_..._! LV LL- .111 Il k,yly Uvil;ltit tJ1 it]t Polar coordinate system to extract texture features. Also, the present Invention relates to a texture-based retrieval method of images indexed by the texture description
method. 10 BACKGROUND OF THE INVENTION
The texture information of an image is one of the most important visual characteristics of the image and thus, has been studied together with the color information for a long time. This texture information of image is usually used as an important low-level visual descriptor in content-based indexing and in abstracting 15 image or video data. Also, Image texture is very important information used for retrieval of a special picture in an electronic album or content-based retrieval in tiles or textiles database.
Up to now, feature values were mostly computed in time domain or in frequency domain to extract a texture feature of the image. More particularly, the 20 method of extracting the texture features in frequency domain was known to be suitable for describing image texture information of various types. Extracting texture features in frequency domain can be done in Cartesian or Polar coordinate system.
Conventionally, the Cartesian coordinate system has been widely used in extracting a texture feature in frequency domain.
25 A paper entitled "Texture Features For Browsing And Retrieval Of Image Data", written by B.S. Manjunath and W.Y. Ma is published in "IEEE Transaction on Pattern Analysis and Machine Intelligence", vol.l8, no.8, in August of 1996, in which a method of dividing the frequency domain of the Cartesian coordinate
( system based on HVS (Human Visual System), filtering an image in the respective channels by Gabor filters, and then extracting the average and the standard deviation from the respective channels as texture features of the image was disclosed.
However, the method of describing image texture is not suitable in the 5 frequency domain of the Cartesian coordinate system for the HVS and leads to poor pcrfonnance in relevant texture images.
For solving the nrohirn, of the it. He te.''e dscptivn method in frequency domain of the Cartesian coordinate system, a paper on image texture description method in frequency domain of the Polar coordinate system was published,
10 in which the texture information in the frequency domain is computed in the Cartesian coordinate system.
In the paper entitled "Rotation-invariant Texture Classification using a complete Space Frequency Model", written by B.S. Manjunath and Geoge M. Haley and published in "IEEE Transaction on Pattern Analysis and Machine Intelligence", l 5 ol 8, no.2, in February of 1999, a method of dividing a frequency space of the Polar coordinate system based on HVS (Human Visual System), then extracting 9 feature values using Gabor filter designed to be suitable for respective channels, and describing the image texture using the extracted feature values of all charmers was disclosed.
However, in this method, the same design of a set of Gabor filters is used 20 for extracting different kinds of texture features in the frequency domain..
SUMMARY OF THE INVENTION
Therefore, the present invention is made in order to solve the aforementioned problems.
An objective of the present invention is to provide a texture description
25 method in a frequency domain, suitable for lIVS, in which image texture features are computed and indexed in a frequency domain.
The other objective of the present invention is to provide a texturebased retrieval method by using texture features computed in the frequency domain of the Polar coordinate system, in which similar images in various variations, such as different
rotations or scales or pixel intensity, are retrieved by comparing a query texture descriptor with a data texture descriptor generated by the texture description method by
taking account for such variations thereof.
The above objectives can be accomplished by a texture description
5 method in the frequency domain of the Polar coordinate system according to the present invention. The texture description method comprises a first step of generating a
frequency layout by naltitioninP aria firP(lupncy 4nmain Into a se+ oratu. c c.al-=lcls; a second step of extracting texture feature values of said image from said respective feature channels; and a third step of constituting a texture descriptor in a vector form by 10 using said texture feature values extracted from said respective feature channels in said frequency layout.
It is preferable that said first step is of generating said frequency layout on the basis of HVS (Human Visual System), and that said frequency domain in said first step is that of the Cartesian coordinate system or the Polar coordinate system.
1: It is more preferable that said first step comprises a sub-step of generating different frequency layouts for different types of texture features, that is, each texture feature type for its respective frequency layout.
It is more preferable that said first step comprises a sub-step of assigning significance or priority to the respective channels.
20 Also, it is preferable that said second step comprises a first substep of Radon-transforming said image; a second sub-step of Fouriertransforming said Radon transforrned image; and a third sub-step of extracting said texture feature values of said Fourier-transformed image from said respective feature channels.
It is more preferable that said third sub-step is of extracting at least 25 energy deviation values and/or energy values in said respective feature channels.
Here, it is still more preferable that a frequency layout for obtaining said energy values and a frequency layout for obtaining said energy deviation value is separately prepared for extracting different types of an image texture, and that said frequency layout for obtaining said energy values partitions said frequency domain at 30 intervals of 2J (0 c I < log2(N/2)-l) octave in a radial direction and at intervals of
( 180/dividing resolution' in an angular direction. The frequency layout for obtaining said energy deviation values partitions said frequency domain at the same intervals in a radial direction and at intervals of 'I 80/dividing resolution' in an angular direction.
It is preferable that said third step is of finding out a rotational reference 5 axis of said image by using said image texture information, rotating said frequency layout with reference to said rotational reference axis, and then extracting said image I texture descriptor of cn1cl image mere the rotation 7 e'=A_o axis IS SCt to tan ai1 aAiS; in a radial direction, m which energy or entropy or a periodical component is most distributed by Radon-transforming said Image.
10 Preferably, the third step is of Fourier-transforming said image to find out a radial reference point, normalizing said Fourier-transformed image with reference to said reference point, and then describing said texture descriptor by using said normalized values of said Fouriertransformed image. Here, the radial reference point is set by finding out an arc in which energy or entropy or a periodical component of 15 said Fourier-transformed image apart at the same distance from the origin in said frequency domain is most distributed, and setting a radius of said founded arc as said radial reference point.
It is preferable that the method of describing image texture in a frequency domain according to the present invention further comprises a fourth step of 20 extracting intensity information of said image to add said intensity information to said texture descriptor.
Also, according to the present invention, a computer readable recording media recording a program for realizing a texture description method in a frequency
domain is provided. The program performs a first step of generating a frequency layout 25 by partitioning said frequency domain into a set of feature channels; a second step of extracting texture feature values of said image by Radon-transforming said image in said respective feature channels, Fourier transforming said Radon-transformed irmage, and extracting texture feature values of said Fourier-transformed image from respective feature channels; and a third step of constituting a texture descriptor of said image in
! vector form by using said texture feature values extracted from said respective feature channels. Also, according to the present invention, a method of populating a database with texture descriptors of images is provided. The method comprises a first 5 step of generating a frequency layout by partitioning the frequency domain into a set of feature channels; a second step of extracting texture feature values of said images in I said respective feature channels a thircl stun (of Co!'stit't''ntr the texture desc.piArS Of said images in vector forms by using said texture feature values extracted in said respective feature channels of said frequency layout; and a fourth step of indexing said 10 respective texture descriptors of said images into said database. The first step comprises a first sub-step of generating the respective frequency layouts for texture feature types by partitioning the frequency domain into the respective sets of feature channels; and a second substep of extracting said texture feature values of each type for saicl images in said feature channels of said respective frequency layouts.
15It is preferable that said second-sub step comprises a first step of extracting energy values of a texture feature type for said images in said feature channels of the corresponding frequency layout for said energy feature type; and a second step of extracting energy deviation values of a texture feature type for said images in said feature channels of the corresponding frequency layout for said energy 20 deviation feature type.
Also, it is preferable that said third step comprises a first sub-step of constituting texture descriptors of said images with said energy values and energy deviation values in a vector form; and a second sub-step of adding the mean and standard deviation values of said images into each of said texture descriptors of said 25 images.
More preferably, the second step comprises extracting energy values and energy deviation values as texture features for said images in said feature channels of said frequency layout.
Still more preferably, the first sub-step comprises a step of generating, I 30 based on HVS, at least more than one frequency layouts for which each frequency s
/ layout is used for extracting feature values of each feature type; and a step of assigning significance or priority to respective channels of said frequency layouts.
Also, it is preferable that the second sub-step comprises a step of Radon transforrning the inputted images; a step of Fourier-transforming said Radon 5 transformed image; and a step of extracting feature values from said Fourier- I transformed image in snarl recnertre fAAt,rr-.11 nmo!c' If c_ a' ljV'U4. into step of extracting feature values from said Fouriertransfomed image is of extracting at least energy values or energy deviation values from said respective feature channels of said frequency layout.
10 Also, a method of retrieving relevant texture images in a database similar to a query image according to the present invention is provided. The method comprises a first step of generating a frequency layout by partitioning frequency domain into a set of feature channels for feature extraction of an input query image; a second step of extracting a query texture descriptor of said query image when said query image is 15 inputted; a third step of readin, a texture descriptor from said database; a fourth step of measuring a distance between said texture descriptor of said input texture image and said texture descriptor read from said database; a fifth step of measuring distances for said texture descriptor of said input image to all or parts of texture descriptors in said database; and a sixth step of ordering the similarity for the said texture descriptor to 20 said texture descriptors in said database using said measured distances.
It is preferable that when rotation-invariant matching of said image is considered, said firm step comprises a first sub-step of measuring distances between a texture descriptor taken from said database and said query texture descriptor by shifting feature values of said query texture descriptor in angular directions into the 25 corresponding positions where the shifted feature values are supposed to extracted when said query image rotates; a second sub-step of measuring the distances between said texture descriptor of said input texture image to said texture descriptor stored in said database for all rotation angles; and a third sub-step of determining as said distance I
! the minimum distance between said texture descriptor of said input texture image and said texture descriptor stored in said database for all rotation angles.
Also, it is preferable that when scale-invariant matching of said image is considered, said fifth step comprises a first sub-step of Donning at least one zoom-in 5 image andior zoom-out image from said query image and extracting said query texture descriptors of zoom-in and/or zoom-out images of said query image; a second sub-step I of measuring the distances between Girl e te!re descriptors Of cv...-i.. I'ur zoom-out query images and said data texture descriptor in said database; and a third sub-step of determining as the distance the minimum distance of said distances behvcen 10 said texture descriptor in said database and said texture descriptor of said query texture descriptors at different scale values. Here, it is preferable that said query texture descriptor and said texture descriptor in said database include a rotational reference axis, a radial reference point, and mean and stand deviation of texture image intensities, respectively. 15It is still more preferable that when rotation-invariant of said query texture descriptor is considered. said fifth step is of aligning said texture descriptor of said query image and said texture descriptor in said database with reference to given rotation angles.
Also, it is preferable that said rotational reference axes are set to be 20 radial axes in which an energy, an entropy or a periodical component is most distributed in Fourier transform of said Radon-transformed images.
Preferably, when intensity-invariant matching of said query texture descriptor is considered, said fifth step is of excluding a mean values from said query texture descriptor and said texture descriptor in said database and measuring a distance 25 between said two texture descriptors.
More preferably, when scale-invariant matching of said query texture image is considered, said fifth step comprises a first sub-step of merging said feature values of the adjacent channels in radial directions for said two texture descriptors to be compared or shifting feature values of said two texture descriptors into radial directions I 30 according to a radial reference point; and a second sub-step of measuring a distance
between said two texture descriptors with feature values merged in adjacent feature channels or with feature values shifted into adjacent feature channels.
Here, said radial reference point is preferably set by finding out an arc in which energy or entropy or periodical components of said Fouriertransfolmed image 5 apart at the same distance from the origin in said frequency domain are most distributed and setting a radius of said foundout arc as said radial reference point.
S t i 1 1 m 0 r e n r p f a r ?,, _ _ _ _...,.. ', a,,, a; i O i I i i I v l i l I i matching is considered simultaneously, said fifth step comprises a first sub-step of merging said feature values of the adjacent channels in radial directions for said two 10 texture descriptors to be compared or shifting feature values of said two texture descriptors into radial directions with reference to a radial reference point; a second sub-step of shifting feature values of said two texture descriptors in angular directions into the corresponding positions where the shifted feature values are supposed to extracted with reference to a rotation point; and a third sub-step of measuring a distance 15 between said two texture descriptors with feature values of adjacent feature channels merged in radial directions and then shifted in annular directions.
A computer readable recording media recording a program retrieving a data image similar to any query image in a computer according to the present invention is provided. The program performs the following steps; a first step of generating a 20 frequency layout by partitioning the frequency domain into a set of feature channels; a second step of, when images to be stored in a database is given, extracting texture feature values of said data image in said respective feature channels, and then extracting and storing a data texture descriptor of said data image by using said extracted texture feature values; a third step of, when said query image is inputted, extracting texture 25 feature values of said query image in said respective feature channels, and extracting a query texture descriptor of said query image by using said extracted texture feature values; a fourth step of matching said data texture descriptor with said query texture descriptor and measuring a distance between two texture descriptors; and a fifth step of determining a similarity between said two images by means of said distance between 30 said two texture descriptors.
Also, a texture-based retrieval method of a data image similar to a query image in a frequency domain according to the present invention is provided. The method comprises a first step of extracting and storing a texture descriptor including texture feature values and the rotation information of images to be stored in a database; 5 a second step of extracting a query texture descriptor including texture feature values and the rotation information of said query image when said query image its inn,.,tted; a thirty CtF'n of 11 (T=[T._ T!.,'.o..f,! wt V. w -;"i Ma. x.a. unapt;|Jiot arli saia query texture descriptor according to said rotation information of said two texture descriptors; a fourth step of matching said two texture descriptors and measuring a 10 distance between said two texture descriptors with rotation angles aligned between said two texture descriptors; and a fifth step of determining a similarity between said two i images by means of said distance between said two texture descriptors.
It is preferable that said step of extracting said texture descriptor in said first and second steps comprises a first sub-step of generating a frequency layout by 15 partitioning the frequency domain into a set of feature channels so as to extract respective feature value: a second substep of extracting texture feature values of said i images in said respective divided frequency domains; and a third sub-step of constituting a texture descriptor of said image in a vector form by using said feature values extracted in said respective frequency channels of said frequency layout.
20 It is more preferable that said step of extracting said rotation information of said images in said first and second steps comprises a first sub-step of finding out a direction in which energy is much distributed in the Fourier transform of said inputted image; a second substep of generating a frequency layout by using said direction as a reference axis; and a third sub-step of adding said rotation information of said 25 frequency layout to said texture descriptor of said image.
It is still more preferable that said first sub-step in said step of extracting; said texture descriptor comprises a step of generating at least one frequency layout in consideration of HVS; and a step of giving significance or priority to respective feature channels of said frequency layouts.
i Preferably, said second sub-step in said step of extracting said texture descriptor comprises a step of Radon-transforming said inputted image; a step of Fourier-transforming said Radon-transformed image; and a step of extracting said texture feature values from said Fourier- transfoTmed image with respect to said 5 respective frequency layout, and it is preferable that the step of extracting texture feature values from said Fourier-transformed image is of extracting at least energy vl',ec or=nerof Eon... '.^v..- Camp v,.
A computer readable recording media recording a program retrieving a data image similar to a query image in a computer according to the present invention is 10 provided. The program performs a first step of generating a frequency layout by partitioning a frequency domain into a set of feature channels; a second step of generating and storing a data texture descriptor by extracting texture feature values and the rotation information of said data image from said respective feature channels when an image to be stored in a database is given; a third step of generating a query texture 15 descriptor by extracting texture feature values and the rotation information of said query image from said respective feature chaMels when said quer image is inputted; a fourth step of aligning the rotating angles between said two data texture descriptors by using said rotation information of said data texture descriptor and said rotation information of said query texture descriptor; a fifth step of matching said two texture descriptors and 20 measuring a distance between said two texture descriptors with said rotating angles aligned between said two texture descriptors; and a sixth step of determining a similarity between said two images by means of said distance between said two texture descriptors. BRIEF DESCRIPTION OF THE DRAWINGS
25 The embodiments of the present invention will be explained with reference to the accompanying drawings, in which: Fig.l is a flow chart for illustrating a texture description method in a
frequency domain according to an embodiment of the present invention; Fig. 2 is a drawing for illustrating general Central Slice theorem;
Fig.3 is a drawing for illustrating a frequency sampling structure in frequency domain using Radon transformation; Fig.4 is a drawing for showing a frequency layout used to extracting average values in respective frequency channels in the present invention; and 5 Fig.S is a drawing for showing a frequency layout used to extracting energy deviation values in respective frequency channels in the present invention.
DETAILED DESCRIPTION OF THE INVENTION
The above objectives, other objectives, features and advantages of the present invention will be better understood from the following description in
10 conjunction with the attached drawings.
Now, an embodiment of the present invention will be described in detail with reference to the drawings.
Fig.] shows a flow chart for illustrating a texture description method in a
frequency domain according to the present invention, using Radon transformation.
I S The texture description method is used both in texture informationbased
indexing of the image and in texture information-based retrieval of the image, in which the input image is processed to prepare a texture descriptor. When images to be stored in a database are given, the corresponding data texture descriptors are generated and the generated texture descriptors are stored in the database. Also, when a query image is 20 inputted, a query texture descriptor is generated and compared with the data texture descriptors stored in the database to perform retrieval.
With reference to Fig. 1, the texture description method according to the
present invention will be described as follows.
First, when any image is inputted (S11), the inputted image is Radon 25 transformed at step S12. Here, Radon-transformation means a serial procedure of performing line integral of 2-dimensional (2-D) image or multi-dimensional multimedia data along a light axis to obtain ldimensional projection data. That is' an object appears different according to viewing angles, and when viewing the object from all
( angles, profiles of the object can be guessed, the Radon transformation is using such principle. The Radon transfonnaiion equation of the 2dimensional image is expressed as follows.
5 [Equation I] p(R) = t<R at f (x y,)dl - f ( f< r >)$(.'cos j i,:. --- s) -
Here,f(x,y) is an image in the Cartesian coordinate system, end po(R) is an l-D projection obtained by the line integration of the image along a light axis of which the angle with respect to a positive x- axis is and which passes through the origin in the Cartesian coordinate system. That is, piR) is an 1-D projection of the 10 image by Radon transformation.
A function x) is a function which becomes 1 when x value is 0. The 2 dimensional image has the range of '-oo < x,y < so' in the Cartesian coordinate system and a range of 'O < s < co, O < < in a Radon coordinate system. That is, when xcos0+y.sind is v, b(xcos0,sin0-s) becomes I. 15 A set of the first Radon transform functions p/R) is referred to as Signogram, and in next step S 13, the Signogram is Fourier transformed. As a result of Fourier transforming the Signogram, relationships between the Fourier transform of the Signogram and the Fourier transform of the image in the Cartesian coordinate system is expressed as following Equation 2.
20 [Equation 2] Go (I) = F(/ cos 0, sin 0) = F(a)X, 6)))!=ACosara}-45ln Here, G Ad) is a function to which pd<R) is Fourier transformed. And,1 is:mx2 +my and this tan(oy/o:) Fig.2 shows a Central Slice theorem and is a drawing for illustrating a relationship between an 1-dimensional Fourier transform of the Signogram and the 25 Signogram. The Fourier transform of the Signogram is a function value taken by cutting the Fourier transform function of the 2-dimensional image along 0-axis.
That is, the image function is Founer-transformed after Radon transforming it, as shown in Fig. 2(b), the resulting Fourier transform of the image is represented in the Polar coordinate system, and frequency sampling in the Polar coordinate system Is shown as in Fig. 3.
5 Fig 3 is a drawing for illustrating a frequency sampling structure in frequency domain using Radon transformation. The Fourier transform using the Revlon +.. r A A T A A - ; _ I I a, I i I i I r o; at C^,. vvvS ERIC it oS,I 1Fv Allay a ieuci y uvi coordinate system. This frequency sampling is described such that the density of the frequency sampling is high in low frequency regions and becomes lower from low to 10 high frequency regions.
This sampling structure is well suited for the characteristics that information of general image texture is gathered in the low-to-mid frequency region, and features extracted from this frequency sampling structure represent well the characteristics of image texture.
15 Next, in step S 14, the image texture features are extracted in the frequency domain of the Polar coordinate system having such frequency sampling structure as shown in Fig. 3. At that time, a frequency layout of the Polar coordinate system generated in step S15 is used. The respective partitioned frequency domains are referred to as a feature channel.
20 The frequency layout is partition of the frequency domain on the basis of HVS (Human Visual System). That is, HVS is shown to be insensitive to the high frequency components and sensitive to the low frequency components of images and the frequency layout is designed by using such characteristics. The details thereof will be described later.
25 The present invention employs respective frequency layouts, that is,energy values and energy deviation values of Fourier transform of the image in respective channels. as an image texture feature. For this reason, a frequency layout of the Polar coordinate system for extracting the energy values and a frequency layout of the Polar coordinate system for extracting the energy deviation values are separately 30 generated.
Fig. 4 is a drawing showing a frequency layout of the Polar coordinate system used to extracting the energy values of respective channels on the basis of HVS.
As shown in Fig. 4, the frequency domain of the Polar coordinate system is partitioned in a radial direction and in an angular direction. The frequency domain is 5 partitioned at intervals of 2' (0 < I < log2(N/2)1) octave in the radial direction and t} is partitioned at intervals of 'IS,O/dividing resolution' in the angular direction Rv this "I.,.iol., a lluciy landau. al <; rOai uuilia;o yicill tU1 CAildLii]g Lilt energy values is dense at low frequency regions and sparse at high frequency regions. The respective partitioned frequency regions indicate feature channels and the slashed part l O is a 5-th channel.
From the above description, a primary characteristic of the present
invention is known, in which the sampling density at low frequency region is high and the sampling density at high frequency region is low due to the Radon transform. When partitioning the frequency domain on the basis of HVS, the low frequency region is 15 partitioned densely and the high frequency region is partitioned sparsely. That is, the feature values extracted from the respective partitioned frequency regions. that is, the respective channels, reflect well the global texture features all together.
Fig. 5 is a drawing for showing a frequency layout used to extract energy deviation values on the basis of HVS.
20 Unlike the frequency layout of the Polar coordinate system for extracting the energy values, the frequency layout of the Polar coordinate system for extracting the energy deviation values uniformly partitions the frequency domain in a radial direction.
However, dis partitioned by 180/P (here, P is a dividing resolution of tI) in the angular direction as in the frequency layout of Fig. 4. The respective partitioned frequency 25 regions constitute feature channels, and the 35-th channel is slashed.
In the present invention, the respective frequency layouts are designed for means of the extracted feature values. This provides a flexibility, so that the optimal frequency layout is allowed to provide high retrieval rate of relevant texture images to the respective features.
! When the energy values and the energy deviation values are obtained in the respective channels, the image texture descriptor describing the image texture from the feature values, that is, a feature vector, is computed in step S16.
The texture descriptor is expressed as following Equation 3 5 [Equation 3] TD=TeA emen,, dot; d,W,{. OPT-} revere, e, is one energy value of the 'th channel m the frequency layout shown in Fig. 4 and dj is the energy deviation value of the -th channel in the frequency layout shown in Fig. 5. Specifically, en represents the energy of a DC channel. P is the number of the frequency regions partitioned in the angular direction and Q is the 10 number of the frequency regions partitioned in the radial direction, in the frequency domain of the Polar coordinate system.
The respective feature values of equation 3 can be first described according to the priority of the channels, and the size of texture descriptor decreases with excluding the feature values of the channels having low significance according to 1 S the significance of channels.
The energy value e, and the energy deviation value dj are obtained by means of Equation 5 and Equation 7, respectively. In Equation 4, p, is obtained by using Gab) which is the Fourier transform of po(R), and in equation 6, qJ IS obtained by using OR() end p, obtained in Equation 4 20 Equation 4] p, = C(i,, 0,)G,j (I) a, [Equation 5] ei =log(1+ p/) [Equation 6] 1. arm. q, L', 4,, a,) j j P. | [Equation 7]
! d, = log(l + q,) As described above, a texture descriptor constituted with the energy values and the energy deviation values of respective feature channels is obtained.
With respect to all the inputted images, step Sll through step S16 are repeatedly performed and the respective data texture descriptors are stored in the 5 database. T he data texture descriptors stored m the database are matched with the query texture descriptor obtained from the query image to be used for retrieval the image similar to the query image.
Hereinafter, the texture-based retrieval method in the Polar coordinate 10 frequency domain will be described.
In order to retrieve the image similar to the query image, 3 elements are considered in the present invention. First, the intensity-invariant matching is considered That is, there are two cases, the one that the images similar in texture are retrieved without considering the changes in intensity of image and with considering the 15 intensity changes. Second, rotation-invariant matching is considered. That is, the retrieval in consideration of rotation of image and the retrieval without consideration of rotation of image is classified. Third, scale-invariant matching is considered. That is, the original image is zoomed in/zoomed out to be retrieved in cases of abridgement/enlargement of the image 20 First, the intensity-invariant retrieval method of texture images is explained. The intensity of image is represented by means of energy value en of the DC channel of the texture descriptor (TD) vector. That is, en is large when the image is bright and indicates small values when the image is dark. Therefore, in the intensity invariant retrieval, en is excluded from TD vector of the data texture descriptor vector 25 and then the TD vector is matched with the query texture descriptor during the similarity matching. However, when the retrieval in consideration of intensity-invariant matching is intended to perform, the TD vector containing en is matched with the query texture descriptor.
! Next, a first embodiment of the retrieval method with invariability in rotation is explained. When the image is rotated with respect to the same image, the conventional texture-based retrieval method did not retrieve the image as the same image. However, in the present invention, by performing matching of the images with 5 invariability in rotation of image, the retrieval may be performed without consideration of rotation. The rotation-invariant retrieval method is as follows.. It is known that a rntnte(1 image In time Our. I in id,= "Ci v'L.;C "rsrvll w1 ÀIIC ully, ial image. In a state that the data texture descriptors TDm are stored in the database, 10 the query image is processed by means of the texture description method of Fig.l to
obtain the query texture descriptor TDqucry Then, a similarity between ary TDm and TDque is computed to measure the matching degree.
The similarity is in inverse proportion to Dm obtained by means of Equation 8.
IS [Equation 8] Dm dis tan ce(TD=, TDquery) A distance between the data texture descriptor and the query texture descriptor is obtained by comparing the texture descriptor having energy and energy deviation values. As explained above, the result that any image has been rotated and then Fourier transformed is equal to the result that the image has been Fourier 20 transformed and then rotated in the frequency domain. When two images are compared while rotating them in the frequency domain, two similar images can be found out.
Therefore, in the present invention, in comparing the distance between two texture descriptors by comparing the texture descriptors, the matching is performed in consideration of possibility of rotation. By that consideration, all the rotated similar 25 images can be retrieved. The matching is represented as following Equation 9.
[F.aatinn 91 . Di' = di.v tan cef TDm It' TDquery)
Here, is 1 801P, and k is any integer between 1 and P. That is, Equation 9 is the equation for obtaining the distance between the rotated data texture descriptor and the query texture descriptor, with the data texture descriptor rotated by the angle in the frequency domain.
5 By applying the distances in respective rotational angle ranges obtained in Equation 9 to following Founfion 10, the minimum. distance is found out.
=qua.iol, I] Dm = min(Dm)|keil...P} By comparing the data texture descriptor with the query texture descriptor with the data texture descriptor rotated by a minute angle and selecting the 10 minimum distance between two texture descriptors as a distance between two texture descriptors, the similar image can be retrieved regardless of rotation of image. On the contrary, when the retrieval without considering invariability in rotation of image is performed, the similarity is retrieved by means of Equation 8.
In retrieval in consideration of invariability in rotation of image, as 15 described above, en is contained in the texture descriptor vector when the intensity of image is considered, and en is excluded from the texture descriptor vector when the intensity of image is not considered. Thereafter, texture-based retrieval in consideration of invariability in rotation is performed.
Now, a second embodiment of the retrieval method with invariability in 20 rotation of image is explained. As described above, Fourier transform of the rotated image is equal to the result of rotating Fourier transform of the non-rotated image in the frequency domain. Therefore, in matching the texture descriptors, when the matching is performed in consideration of possibility of rotation, all images having equal texture and being rotated can be retrieved. For this performance, in the first embodiment of the 25 retrieval method with invariability in rotation, the method of matching the data texture descriptors with the query texture descriptor is provided, with the data texture descriptor rotated by a minute angle.
On the contrary, in the second embodiment of the present invention, a method of adding the rotation information to the texture dcscuptor is provided. That is, if the image is Radon transformed, a reference angular direction which is most periodical or in which energy is most distributed is known. The Polar coordinate S frequency layout (transformed layout) is generated using the direction as a reference axis, and then the texture descriptor of Fi=. 1 is c^rn.puted. At that tinge, the irabfvrrrev layout is rotated with respect to line Polar coordinate layout of Fig. A, and the reference axis of the transformed layout and the reference axis of the original Polar coordinate frequency layout are added to the texture descriptor as rotation information.
10 When a retrieval is required while the query texture descriptor containing the rotation information is provide together with the database storing the data texture descriptors containing the rotation information, two texture descriptors are matched in the rotational angles by using the rotation information of the data texture descriptor and the rotation of the query texture descriptor.
15 That is, the reference angular directions of two images are matched and in this state, the distance between two images is obtained by comparing two texture descriptors. Unlike the first embodiment, the second embodiment has an advantage that similarity between two images can be obtained without the procedure of obtaining the distance between two texture descriptors by comparing the data texture descriptor with 20 the query texture descriptor while the data texture descriptor is rotated. However, because the procedure of obtaining and adding the rotation information to the texture descriptor is added to the step of describing the texture, computing the texture descriptor become complex.
Here, the rotation information is represented using Radon transform of 25 image, in which the reference direction is the direction in which energy is most periodic or in which energy is most distributed. However, a method of finding out the reference ..; A, _;A1 hi; - A; 't=; the tAVtlTrA in{AOti" of imp And Pcrrihin the.
HA l V A V A.. D - D - - - V texture by matching the frequency layout with the reference axis, or a texture description method using the frequency layout without the reference axis may be
30 employed and the present invention is not limited to those method.
Third, in the texture-based retrieval method of an embodiment according to the present invention, as described above, invariability in abridgementlenlargement of image is considered, and is explained in detailed.
When image is obtained as varying zoom of a camera, the obtained 5 image is abridged or enlarged according to the zoom magnification of the camera.
When such effect is analyzed in the frequency domain, the frequency spectrum.
distiutiGn Of an,i,,aE,e ell Bilged fiend ill; i;hla; image srwinKs toward one origin of the frequency domain than the original spectrum. Also, the frequency spectrum distribution of an image abridged from the original image spreads out from the origin of lO the frequency domain than the original spectrum.
By Radon transformation of image, scale reference is found out with reference to energy of the projection data. When the texture-based retrieval with invariability in abridgement/enlargement of image is performed with respect to such image, by adding feature value of adjacent channel with reference to the scale reference 15 in a radial direction to be overlapped by one channel, or by finding out a channel enlarged/abridged from the origin due to abridgement/enlargement, the similarity is computed as in Equation 11. The added channel is referred to as an merged channel or a matching channel and in a word, as a modified channel.
[Equation ll] Dk = dis tan ce(modiJied channel featureeXrute, modified channel featuretnb'O, n) 20 When the texture descriptor is obtained by finding out a reference point in the radial direction by using texture information of image, normalizing Fourier transform of the image with reference to the reference point, and then extracting feature value of the normalized Fourier transform, similarity retrieval can be performed using Equation 8.
25 Here, a radius of the found-out arc as follows is set as the reference point in the radial direction. The arc is found out in which energy or entropy or periodical component of said Fourier-transformed image apart at the same distance from the origin in the frequency domain is most distributed is found out.
Another embodiment of the retrieval method with invariability in abridgement/enlargement is described. By making one inputted image at least one enlarged image or at least one abridged image, the texture descriptor is represented by means of respective enlargedlabridged query images. Then, the texture descriptors of 5 the respective enlarged/abndged query images are extracted and a distance to the data texture descriptor is measured by means of lquatinn The minimum, distance of the istn_c eL-Y-iCCn tale data descriptui Gnu ills Uris descriptors OI one respective enlarged/abridged query image is determined as an original distance between the query texture descriptor and the data texture descriptor.
10 When the retrieval without invariability in abridgement/enlargement of image is performed, the similarity is computed by means of Equation 8.
When the retrieval with both invariability in abridgement/enlargement and invariability in rotation is performed, in the retrieval with invariability in abridgementlenlargement of image, adjacent channels are merged and modified, and 15 then feature values of the modified channels are retrieved invariantly in rotation. At that time, the similarity retrieval is performed by means of Equations 1 1, 9 and 10.
The texture descriptor TD vector expressed in equation 3 makes the texture structure information be inferred from the arranged pattern of feature values of the texture descriptor. This can supports for the functionality of roughly finding out a 20 special structure of the texture to browse In order to support the simple browsing, in the present invention, the simple structure information of texture is computed using feature values extracted from the energy channel as show in Fig. 4. Computing the texture structure is performed by obtaining entropy of feature values of 30 energy channels, or by computing the angular difference or radial difference between two 25 maximum energies in energy charnel layout.
According to the above-described present invention, by using a method of nrtitirning the fren'ncv rlomain in the Polar coordinate system with a frequency layout in the Polar coordinate system suitable for extracting the respective feature values, a method of extracting feature values in respective frequency domains, 30 assigning significance and priority to respective frequency channels, a texture indexing
method supporting rotation-, scale-, intensity-invariant retrieval, a texture descriptor matching method, and the like, the image texture can be described more accurately and the effective indexing and retrieval is possible.
The image texture descriptor extracted by means of the texture 5 description method according to the present invention can be used as a useful searching
clue in finding out an image having a special feature in an acne! photograph or. a And ale, a Am! a, -a6 > 'I'^ Although preferred embodiments of the present invention has been disclosed with reference to the appended drawings, descriptions in the present
10 specification are only for illustrative purpose, not for limiting the present invention.
Also, those who are skilled in the art will appreciate that various modifications, additions and substitutions are possible without departing from the scope and spirit of the present invention. Therefore, it should be understood that the present invention is not limited only to the accompanying claims and the equivalents thereof, 15 and includes the aforementioned modifications, additions and substitutions.

Claims (3)

  1. ( A method of retrieving relevant texture images in a database similar to
    a query image, comprising the steps of: a first step of generating a frequency layout by partitioning frequency domain into a set of feature channels for feature extraction of an input query image; a second step of extracting a query texture descriptor of said query image when said query image is inputted; a third step of reading a textu1 c dCSC'iptvl from said database; a fourth step of measuring a distance between said texture descriptor of said input texture image and said texture descriptor read from said database; a fifth step of measuring distances for said texture descriptor of said input image to all or parts of texture descriptors in said database; and a sixth step of ordering the similarity for the said texture descriptor to said texture descriptors in said database using said measured distances.
  2. 2. By. The method of retrieving relevant texture images in a database similar to a query image according to claim wherein when rotationinvariant matching of said image is considered. said fifth step comprises: a first sub-step of measuring distances between a texture descriptor taken from said database and said query texture descriptor by shifting feature values of said query texture descriptor in angular directions into the corresponding positions where the shifted feature values are supposed to extracted when said query image rotates; a second sub-step of measuring the distances between said texture descriptor of said input texture image to said texture descriptor stored in said database for all rotation angles; and a third sub-step of determining as said distance the minimum distance between said texture descriptor of said input texture image and said texture descriptor stored in said database for all rotation angles.
    ( /
  3. 3.. The method of retrieving relevant texture images in a database similar to a query image according to claim Ad, wherein when scaleinvariant matching of said image is considered, said fifth step comprises: a first sub-step of forming at least one zoom-in image andlor zoom-out image from said query image and extracting said query texture descriptors of zoom-in anchor zoom out images of said query image; a scccl,d sub-step Or measuring tile distances between said query texture descriptors of zoom-n and/or zoom-out query images and said data texture descriptor in said database; and a third sub-step of determining as the distance the minimum distance of said distances between said texture descriptor in said database and said texture descriptor of said query texture descriptors at different scale values.
    The method of retrieving relevant texture images in a database similar to a query image according to claim I, wherein said query texture descriptor and said texture descriptor in said database include a rotational reference axis, a radial reference point, and mean and stand deviation of texture image intensities, respectively.
    5 I. The method of retrieving relevant texture images in a database similar to a query image according to claim wherein when rotationinvariant of said query texture descriptor is considered, said fifth step is of aligning said texture descriptor of said query image and said texture descriptor in said database with reference to given rotation angles.
    The method of retrieving relevant texture images in a database similar to a query image according to claim I, wherein said rotational reference axes are set to be radial axes in which an energy, an entropy or a periodical component is most distributed in Founer transform of said Radon-transformed images 7 Q. The method of retrieving relevant texture images in a database similar to a query image according to claim At, wherein when intensity-invariant matching of said query texture descriptor is considered, said fifth step is of excluding a mean values from said query
    if/ texture descriptor and said texture descriptor in said database and measuring a distance between said two texture descriptors.
    my. The method of retrieving relevant texture images in a database similar to a query image according to claim wherein when scale-invariant matching of said query texture image is considered, said fifth step comprises: a first sub-stc;p of.leril.g said fGa.-ui-e -values of the aujaceilc cl-aru-els ill radial directions for said two texture descriptors to be compared or shifting feature values of said two texture descriptors into radial directions according to a radial reference point; and a second sub-step of measuring a distance between said two texture I descriptors with feature values merged in adjacent feature channels or with feature values shifted into adjacent feature channels.
    9 The method of retrieving relevant texture images in a database similar to a query image according to claims, wherein said radial reference point is set by fmding out an arc in which enerYv or entropy or periodical components of said Fourier-transfonned image apart at the same distance Mom the origin in said frequency domain are most distributed and setting a radius of said found-out arc as said radial reference point.
    10 Ad. The method of retrieving relevant texture images in a database similar to a query image according to claim 7 wherein when scaleinvariant and rotation-invariant matching is considered simultaneously, said fifth step comprises: a first sub-step of merging said feature values of the adjacent channels in radial directions for said two texture descriptors to be compared or shifting feature values of said two texture descriptors into radial directions with reference to a radial reference point; a second sub-step of shifting feature values of said two texture descriptors in angular directions into the corresponding positions where the shifted feature values are supposed to extraclea with refr;u; in a Rotation swine; -,-
    a third sub-step of measuring a distance between said two texture descriptors with feature values of adjacent feature channels merged in radial directions and then shifted in angular directions.
    l 1 Ad; A computer readable recording media recording a program retrieving a data image similar to any query image in a computer, the program performing the steps of: a first step of generating a frequency layout by partitioning the tTequency domain into a set of feature channels; a second step of, when images to be stored in a database is given, extracting texture feature values of said data image in said respective feature channels, and then ey.tracting and st^n.e, a data tcx.- a.c dcsvllptvl of said data image 'oy using said extracted texture feature values; a third step of, when said query image is inputted, extracting texture feature values of said query image in said respective feature channels, and extracting a query texture I descriptor of said query image by using said extracted texture feature values; a fourth step of matching said data texture descriptor with said query texture descriptor and measuring a distance between two texture descriptors; and a fifth step of determining a similarity between said two images by means of said distance between said two texture descriptors.
    1 2 texture-based retries al method of a data image similar to a query image in a Dequency domain, comprising the steps of: a first step of extracting and storing a texture descriptor including texture feature values and the rotation information of images to be stored in a database; a second step of extracting a query texture descriptor including texture feature values and the rotation information of said query image when said query image is inputted; a third step of aligning the rotating angle between said data texture descriptor and said query texture descriptor according to said rotation information of said two texture descriptors; a fourth step of matching said two texture descriptors and measuring a distance between said two texture descriptors with rotation angles aligned between said two texture descriptors; and a fifth step of determining a similarity between said two images by means of said distance between said two texture descriptors.
    ( ! The texture-based retrieval method of a data image similar to a query image in a frequency domain according to claim By, wherein said step of extracting said texture descriptor in said first and second steps comprises: a first sub-step of generating a frequency layout by partitioning the frequency domain into a set of feature channels so as to extract respective feature value; a second sub-step of extracting texture feature values of said images in said respective diviler1 fren,,er!cy rlomains; and iiiru suD-siep of consuming a texture descriptor of said image in a vector fond by using said feature values extracted in said respective frequency channels of said frequency layout. I $ The texturebased retrieval method of a data image similar to a query image in a frequency domain according to claim 3; wherein said step of extracting said rotation information of said images in said first and second steps comprises: a first sub-step of finding out a direction in which energy is much distributed in the Fourier transform of said inputted image; a second sub-step of generating a frequency layout by using said direction as a reference axis; and a third sub-step of adding said rotation information of said frequency layout to said texture descriptor of said image.
    )538. The texture-based retrieval method of a data image similar to a query image in a frequency domain according to claim <, wherein said first sub-step in said step of extracting said texture descriptor comprises: generating at least one frequency layout in consideration of HVS; and giving significance or priority to respective feature channels of said frequency layouts.
    1:, Ad. The texture-based retrieval method of a data image similar to a query image in a frequency domain according to claim wherein said second sub-step in said step of extracting said texture descriptor comprises: Radon-transforming said inputted image; Fourier-transfonT ing said Radontransformed image; and extracting said texture feature values from said Fourier-transformed image Ah respect to said.espcctnve cqucr.cy'ayout.
    17 I The texture-based retrieval method of a data image similar to a query image in a frequercy domain according to claim wherein said step of extracting texture feature values from said Fourier-transformed image is of extracting at least energy values or energy deviation values in said respective feature channels.
    18 A computer readable recording media recording a program retrieving a data image similar to a query image in a computer, the program performing steps of: a first step of generating a frequency layout by partitioning a frequency domain into a set of feature channels: a second step of generating and storing a data texture descriptor by extracting texture feature values and the rotation information of said data image from said respective feature channels when an image to be stored in a database is given; a third step of generating a query texture descriptor by extracting texture feature values and the rotation information of said query image from said respective feature channels when said query image is inputted; a fourth step of aligning the rotating angles between said two data texture descriptors by using said rotation information of said data texture descriptor and said rotation information of said query texture descriptor; a fifth step of matching said two texture descriptors and measuring a distance between said two texture descriptors with said rotating angles aligned between said two texture descriptors; and a sixth step of determining a similarity between said two images by means of said distance between said two texture descriptors.
    rFk9 rle\laWt texture I6A(S ivy We \9 Ad,- A method of dercnbR image texture information in a frequency domain substantially as described herein with reference to the accompanying drawings.
    2 at Apparatus substantially as described herein with reference to the accompanying drawings. Z9
GB0319202A 1999-07-09 2000-07-06 Texture description method and texture-based retrieval method in frequency domain Expired - Fee Related GB2388761B (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
KR19990027805 1999-07-09
KR1019990041574A KR100308456B1 (en) 1999-07-09 1999-09-28 Texture description method and texture retrieval method in frequency space
GB0016488A GB2355378B (en) 1999-07-09 2000-07-06 Texture description method and texture-based retrieval method in frequency domain

Publications (3)

Publication Number Publication Date
GB0319202D0 GB0319202D0 (en) 2003-09-17
GB2388761A true GB2388761A (en) 2003-11-19
GB2388761B GB2388761B (en) 2004-01-21

Family

ID=29254907

Family Applications (2)

Application Number Title Priority Date Filing Date
GB0325991A Expired - Fee Related GB2392073B (en) 1999-07-09 2000-07-06 Texture description method and texture-based retrieval method in frequency domain
GB0319202A Expired - Fee Related GB2388761B (en) 1999-07-09 2000-07-06 Texture description method and texture-based retrieval method in frequency domain

Family Applications Before (1)

Application Number Title Priority Date Filing Date
GB0325991A Expired - Fee Related GB2392073B (en) 1999-07-09 2000-07-06 Texture description method and texture-based retrieval method in frequency domain

Country Status (1)

Country Link
GB (2) GB2392073B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130121570A1 (en) * 2011-09-17 2013-05-16 Zhe Lin Methods and Apparatus for Visual Search
US8874557B2 (en) 2011-09-02 2014-10-28 Adobe Systems Incorporated Object retrieval and localization using a spatially-constrained similarity model

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7240075B1 (en) * 2002-09-24 2007-07-03 Exphand, Inc. Interactive generating query related to telestrator data designating at least a portion of the still image frame and data identifying a user is generated from the user designating a selected region on the display screen, transmitting the query to the remote information system
US8880563B2 (en) 2012-09-21 2014-11-04 Adobe Systems Incorporated Image search by query object segmentation

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5465308A (en) * 1990-06-04 1995-11-07 Datron/Transoc, Inc. Pattern recognition system
WO2000023918A1 (en) * 1998-10-19 2000-04-27 Sony Electronics, Inc. Modular image query system
US6192150B1 (en) * 1998-11-16 2001-02-20 National University Of Singapore Invariant texture matching method for image retrieval
US20020099721A1 (en) * 1999-01-25 2002-07-25 Lucent Technologies Inc. Retrieval and matching of color patterns based on a predetermined vocabulary and grammar

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100355404B1 (en) * 1999-12-03 2002-10-11 삼성전자 주식회사 Texture description method and texture retrieval method using Gabor filter in frequency domain

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5465308A (en) * 1990-06-04 1995-11-07 Datron/Transoc, Inc. Pattern recognition system
WO2000023918A1 (en) * 1998-10-19 2000-04-27 Sony Electronics, Inc. Modular image query system
US6192150B1 (en) * 1998-11-16 2001-02-20 National University Of Singapore Invariant texture matching method for image retrieval
US20020099721A1 (en) * 1999-01-25 2002-07-25 Lucent Technologies Inc. Retrieval and matching of color patterns based on a predetermined vocabulary and grammar

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8874557B2 (en) 2011-09-02 2014-10-28 Adobe Systems Incorporated Object retrieval and localization using a spatially-constrained similarity model
US8983940B2 (en) 2011-09-02 2015-03-17 Adobe Systems Incorporated K-nearest neighbor re-ranking
US20130121570A1 (en) * 2011-09-17 2013-05-16 Zhe Lin Methods and Apparatus for Visual Search
US8781255B2 (en) 2011-09-17 2014-07-15 Adobe Systems Incorporated Methods and apparatus for visual search
US8805116B2 (en) * 2011-09-17 2014-08-12 Adobe Systems Incorporated Methods and apparatus for visual search

Also Published As

Publication number Publication date
GB2388761B (en) 2004-01-21
GB2392073B (en) 2004-03-31
GB2392073A (en) 2004-02-18
GB0325991D0 (en) 2003-12-10
GB0319202D0 (en) 2003-09-17

Similar Documents

Publication Publication Date Title
CA2367966C (en) Texture description method and texture-based retrieval method in frequency domain
JP3635368B2 (en) Content-based image search system and search method thereof
JP3560159B2 (en) Texture description method using Gabor filter in frequency space and texture-based video retrieval method
US6072904A (en) Fast image retrieval using multi-scale edge representation of images
US6804684B2 (en) Method for associating semantic information with multiple images in an image database environment
Bach et al. Virage image search engine: an open framework for image management
US6445834B1 (en) Modular image query system
Karthika et al. Digital video copy detection using steganography frame based fusion techniques
AU2011200696B2 (en) Method, apparatus and system for rating images
US20100030778A1 (en) Retrieving and sharing electronic documents using paper
CN108491543A (en) Image search method, image storage method and image indexing system
GB2388761A (en) Texture-based retrieval of similar images
Nguyen et al. TagNSearch: Searching and navigating geo-referenced collections of photographs
JP2000010989A (en) Similar object retrieval method/device and recording medium recording similar object retrieval program
Raisi et al. Applying content-based image retrieval techniques to provide new services for tourism industry
CN107967300B (en) Method, device and equipment for retrieving organization name and storage medium
WO2020103074A1 (en) Image searching method and apparatus, storage medium, and electronic device
Borowski et al. Structuring the visual content of digital libraries using CBIR systems
Petkovic et al. Content-based retrieval of spatio-temporal video events
Shih et al. Intelligent virtual world: technologies and applications in distributed virtual environment
Mustafa et al. Creating agents for locating images of specific categories
Mlsna Color sets with morphological and B-spline enhancements for content-based image retrieval
CN112990835A (en) Offline storage retrieval method, system, device and medium
RO135778A0 (en) System and method for saving and retrieving certain user-defined regions or reference points in an n-dimensional data set
Naaman Leveraging Geo-Referenced Digital Photographs Thesis Introduction

Legal Events

Date Code Title Description
REG Reference to a national code

Ref country code: HK

Ref legal event code: DE

Ref document number: 1061084

Country of ref document: HK

REG Reference to a national code

Ref country code: HK

Ref legal event code: GR

Ref document number: 1061084

Country of ref document: HK

PCNP Patent ceased through non-payment of renewal fee

Effective date: 20140706