CN105022752A - Image retrieval method and apparatus - Google Patents
Image retrieval method and apparatus Download PDFInfo
- Publication number
- CN105022752A CN105022752A CN201410175999.3A CN201410175999A CN105022752A CN 105022752 A CN105022752 A CN 105022752A CN 201410175999 A CN201410175999 A CN 201410175999A CN 105022752 A CN105022752 A CN 105022752A
- Authority
- CN
- China
- Prior art keywords
- image
- retrieved
- index table
- cluster index
- color
- 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
Links
Abstract
The invention relates to an image retrieval method and apparatus. The method comprises the steps of: obtaining a to-be-retrieved image from a user terminal; converting the to-be-retrieved image to HSV space from RGB space; quantizing the converted to-be-retrieved image into N-dimensional color features; sorting the feature values of the N-dimensional color features in a descending order, and selecting the first M-dimensional color features as main color of the to-be-retrieved image; determining a cluster index table name of the to-be-retrieved image according to an index value corresponding to the main color; querying whether a cluster index table with the same name as the cluster index table of the to-be-retrieved image exists in the established cluster index table according to a name of the cluster index table; if existence, obtaining an image index of the corresponding image; querying main color percentage of the corresponding image according to the obtained image index; calculating the image similarity according to the main color percentage; and returning an image matched with the to-be-retrieved image to the user terminal according to the similarity. The image retrieval method and apparatus disclosed by the invention remarkably improve the image retrieval efficiency and accuracy.
Description
Technical field
The disclosure relates to searching computer field, especially, relates to a kind of image search method and device.
Background technology
Along with the fast development that the universal of internet, applications and multimedia technology, computer technology are applied, and the popularization and application of mass-memory unit and digitizer, the manifestation mode of information is progressively developed into by traditional text mode based on the form of the multimedia messagess such as figure, image, video, audio frequency.Wherein, the most important carrier that image transmits as information, has been deep into the various aspects of people's life.Therefore, from jumbo image data base, how to find required digital picture rapidly, exactly has become one of focus of multimedia technology research in recent years, has wide economic worth and market outlook.
Color characteristic is as one of the method for image overall feature interpretation, and definition is relatively clearer and more definite, extracts easily, has rotation, translation invariance, and all insensitive to various distortion, shows great robustness.Therefore, the image retrieval based on color characteristic is most widely used, CBIR method.MPEG-7 dominant color descriptors is that the one of MPEG-7 describes the factor, and it instead of the feature of whole image with a small amount of representative color.
At present, can adopt the method for the number percent first calculating the color be at every turn extracted, if number percent unnecessary 5%, corresponding color relation is less than the quantized color of 60% as domain color by as domain color or cumulative number percent.But the quantity of the dominant color descriptors extracted by this method is not fixed, add the complicacy of retrieving, recall precision is low.
The method of fixing domain color quantity can also be adopted, but, so far do not propose a kind of method of quick indexing image, all inquire about and Similarity Measure all images in database when each inquiry, this will reduce effectiveness of retrieval and speed greatly.
Summary of the invention
The disclosure proposes new technical scheme in view of at least one in above problem.
The disclosure provides a kind of image search method in one, and it significantly improves recall precision to image and accuracy.
The disclosure on the other hand provides a kind of image retrieving apparatus at it, and it significantly improves recall precision to image and accuracy.
According to the disclosure, a kind of image search method is provided, comprises:
User terminal obtains image to be retrieved, and image to be retrieved uses rgb space to represent;
Image to be retrieved is transformed into HSV space by rgb space;
Be that N ties up color characteristic by the image quantization to be retrieved being transformed into HSV space;
The N of image to be retrieved is tieed up the eigenwert descending sort by size of color characteristic, choose the front M after sequence and tie up the domain color of color characteristic as image to be retrieved, wherein, M<N;
The index value corresponding according to the domain color of image to be retrieved determines that the cluster index table name of image to be retrieved claims;
Whether exist in the cluster index table set up based on image each in image data base according to the cluster index table name query of the image to be retrieved determined and claim identical cluster index table with the cluster index table name of image to be retrieved;
As existed, then from claiming to obtain image index in identical cluster index table with the cluster index table name of image to be retrieved;
The image mass-tone number percent corresponding according to the image index inquiry obtained;
Based on the similarity between image mass-tone percentage calculation image to be retrieved and the image inquired from image data base;
According to the similarity between image, the image with images match to be retrieved is returned to user terminal.
In embodiments more of the present disclosure, the method also comprises:
Cluster index table is set up based on image each in image data base.
In embodiments more of the present disclosure, the step setting up cluster index table based on image each in image data base comprises:
Each image successively in reads image data storehouse, wherein, each image in image data base uses rgb space to represent;
Each image in image data base is transformed into HSV space by rgb space;
Be that N ties up color characteristic by each image quantization being transformed into HSV space;
The N of each image is tieed up the eigenwert descending sort by size of color characteristic, choose the front M after sequence and tie up the domain color of color characteristic as each image;
Claim according to the cluster index table name of each image in the index value determination image data base that the domain color of each image is corresponding;
The image index value of each image and corresponding mass-tone number percent are saved in corresponding cluster index table.
In embodiments more of the present disclosure, the tone H in HSV space is divided into 9 parts equably, saturation degree S is divided into 3 parts, brightness V is divided into 3 parts.
According to the disclosure, additionally provide a kind of image retrieving apparatus, comprising:
Image acquisition unit, obtains image to be retrieved for user terminal, and image to be retrieved uses rgb space to represent;
Space conversion unit, for being transformed into HSV space by image to be retrieved by rgb space;
Image quantization unit, for by the image quantization to be retrieved being transformed into HSV space being N dimension color characteristic;
Domain color determining unit, for the N of image to be retrieved being tieed up the eigenwert descending sort by size of color characteristic, choosing the front M after sequence and tieing up the domain color of color characteristic as image to be retrieved, wherein, and M<N;
For the index value corresponding according to the domain color of image to be retrieved, concordance list title determining unit, determines that the cluster index table name of image to be retrieved claims;
Judging unit, claims identical cluster index table for whether existing in the cluster index table set up based on image each in image data base according to the cluster index table name query of the image to be retrieved determined with the cluster index table name of image to be retrieved;
Mass-tone number percent acquiring unit, for such as existing, then from claiming to obtain image index in identical cluster index table with the cluster index table name of image to be retrieved, and the image mass-tone number percent corresponding according to the image index inquiry obtained;
Similarity calculated, for based on the similarity between image mass-tone percentage calculation image to be retrieved and the image inquired from image data base;
Result returns unit, for the image with images match to be retrieved being returned to user terminal according to the similarity between image.
In embodiments more of the present disclosure, this device also comprises:
Unit set up by concordance list, for setting up cluster index table based on image each in image data base.
In embodiments more of the present disclosure, concordance list is set up unit and is comprised image acquisition unit, space conversion unit, image quantization unit, domain color determining unit, concordance list title determining unit and storing sub-units, wherein,
Image acquisition unit, also for each image in reads image data storehouse successively, wherein, each image in image data base uses rgb space to represent;
Space conversion unit, also for each image in image data base is transformed into HSV space by rgb space;
Image quantization unit, also for by each image quantization being transformed into HSV space being N dimension color characteristic;
Domain color determining unit, also for the N of each image being tieed up the eigenwert descending sort by size of color characteristic, choosing the front M after sequence and tieing up the domain color of color characteristic as each image;
Concordance list title determining unit, also for claiming according to the cluster index table name of each image in index value determination image data base corresponding to the domain color of each image;
Storing sub-units, for being saved in the image index value of each image and corresponding mass-tone number percent in corresponding cluster index table.
In embodiments more of the present disclosure, the tone H in HSV space is divided into 9 parts by image quantization unit equably, saturation degree S is divided into 3 parts, and brightness V is divided into 3 parts.
In technical scheme of the present disclosure, owing to by the image quantization in HSV space being N dimension, and according to the domain color of eigenwert determination image, the cluster index table name of the domain color determination image of image is utilized to claim, first the image with identical domain color is searched when treating retrieving images and retrieving, but retrieve in the image with identical domain color again, visible, the scope of retrieval is significantly reduced compared with existing image search method.Meanwhile, due to the range of search of the domain color determination image according to image, the accuracy of image retrieval is also further ensured.
Accompanying drawing explanation
Accompanying drawing described herein is used to provide further understanding of the disclosure, forms a application's part.In the accompanying drawings:
Fig. 1 is the schematic flow sheet of the image search method of a disclosure embodiment.
Fig. 2 is the system architecture schematic diagram that the disclosure realizes image retrieval.
Fig. 3 is the schematic diagram that the disclosure is changed by RGB to HSV.
Fig. 4 is the structural representation of the image retrieving apparatus of a disclosure embodiment.
Embodiment
Below with reference to accompanying drawings the disclosure is described.It should be noted that following being described in is only explanatory and exemplary in essence, never as any restriction to the disclosure and application or use.Unless stated otherwise, otherwise positioned opposite and numerical expression and the numerical value of the parts of setting forth in an embodiment and step do not limit the scope of the present disclosure.In addition, technology well known by persons skilled in the art, method and apparatus may not be discussed in detail, but are intended to the part becoming instructions in appropriate circumstances.
The disclosure is in order to overcome the shortcoming and defect of prior art, and provide a kind of cluster image retrieval algorithm efficiently, this algorithm can be applied to the image retrieval to MPEG-7.It adopts the colouring information of HSV (Hue Saturation Lightness, hue, saturation, intensity) color model quantized image, the quantity of fixing domain color.Only need when retrieving image to retrieve the image of specific cluster index table, because the process of establishing of specific cluster index table is exactly the main color characteristic according to image, choose the image composition that similarity is higher, and then avoid the invalid retrieval of the image to a large amount of dissmilarity or low similarity, thus significantly improve validity and the efficiency of retrieval.
Fig. 1 is the schematic flow sheet of the image search method of a disclosure embodiment.
As shown in Figure 1, the flow process in this embodiment can comprise:
S102, user terminal obtains image to be retrieved, and this image to be retrieved uses RGB (Red Green Blue) space representation, such as, MPEG-7 image.
S104, is transformed into HSV space by image to be retrieved by rgb space;
Wherein, in HSV space, H represents tone, and its span is 0 ° ~ 360 °, counterclockwise calculates from 0 °, red with 0 ° of representative, green with 120 ° of representatives, blue with 240 ° of representatives.S represents saturation degree, and its span is 0 ~ 1.V represents brightness, and its span is 0 ~ 1.The concrete existing conversion method that can utilize realizes from rgb space to the conversion of HSV space.
The image quantization to be retrieved being transformed into HSV space is that N ties up color characteristic by S106;
Particularly, inventor finds, the both sides of red (0 °) (such as, 360 ° to 20 °) also close to red, the both sides of green (120 °) (such as, 100 ° to 140 °) also close to green, the both sides (such as, 220 ° to 260 °) of blue (240 °) are also close to blue.Therefore, tone, brightness and saturation degree are divided into multiple interval respectively, and then can are multidimensional by the image quantization in HSV space, to segment the color characteristic of picture of publishing picture further, wherein, N be greater than 0 integer.
S108, ties up the eigenwert descending sort by size of color characteristic by the N of image to be retrieved, choose the front M after sequence and tie up the domain color of color characteristic as image to be retrieved, wherein, M<N, that is, using the domain color of M color characteristic before eigenwert is higher as image.
S110, the index value corresponding according to the domain color of image to be retrieved determines that the cluster index table name of image to be retrieved claims;
Concrete, owing to by image quantization being N dimension, therefore, the combination of each tone quantizing, saturation degree and brightness all can be represented by certain index value in 1 ~ N, this index value secondary indication tone of image, saturation degree and brightness.
When determining that the cluster index table name of image to be retrieved claims, form cluster index table name according to the index order after eigenwert descending sort to claim, that is, both embody the domain color that this image comprises in this title, also embody these domain color ratio in the images simultaneously.
Whether S112, exist in the cluster index table set up based on image each in image data base claim identical cluster index table with the cluster index table name of image to be retrieved according to the cluster index table name query of the image to be retrieved determined;
It may be noted that and be, before image is retrieved, built the cluster index table of each image in image data base.The cluster index table name of the image to be retrieved determined in step S110 is claimed in the cluster index table of each image in reflection image data base, search the cluster index table that whether there is identical cluster index table name and claim, that is, realize searching in image data base the image whether existing and there is with the image that is retrieved identical domain color by searching cluster index table.Visible, the method and existing image one by one carry out, compared with the method for mating, significantly improving recall precision.
S114, as existed, then from claiming to obtain image index in identical cluster index table with the cluster index table name of image to be retrieved.
S116, the image mass-tone number percent corresponding according to the image index inquiry obtained.
S118, based on the similarity between image mass-tone percentage calculation image to be retrieved and the image inquired from image data base.
S120, returns to user terminal according to the similarity between image by the image with images match to be retrieved.
In this embodiment, owing to by the image quantization in HSV space being N dimension, and according to the domain color of eigenwert determination image, the cluster index table name of the domain color determination image of image is utilized to claim, first the image with identical domain color is searched when treating retrieving images and retrieving, but retrieve in the image with identical domain color again, visible, the scope of retrieval is significantly reduced compared with existing image search method.Meanwhile, due to the range of search of the domain color determination image according to image, the accuracy of image retrieval is also further ensured.
Further, before step S102, set up cluster index table based on image each in image data base, to improve the recall precision of image to be retrieved.
Particularly, the step setting up cluster index table based on image each in image data base can comprise:
Each image successively in reads image data storehouse, wherein, each image in image data base uses rgb space to represent;
Each image in image data base is transformed into HSV space by rgb space;
Be that N ties up color characteristic by each image quantization being transformed into HSV space;
The N of each image is tieed up the eigenwert descending sort by size of color characteristic, choose the front M after sequence and tie up the domain color of color characteristic as each image;
Claim according to the cluster index table name of each image in the index value determination image data base that the domain color of each image is corresponding;
The image index value of each image and corresponding mass-tone number percent are saved in corresponding cluster index table.
Wherein, the tone H in HSV space can be divided into 9 parts equably, saturation degree S be divided into 3 parts, brightness V is divided into 3 parts, can image quantization be 81 dimensions, that is, N=81 like this.
In the following embodiments, provide a kind of efficient cluster image retrieval algorithm based on MPEG-7, it is 81 dimensions that application MPEG-7 vision content describes Color Image Quantization, selected characteristic value large front 8 dimension (namely, M=8) as domain color feature, set up cluster index table and database according to domain color combination, utilize the similarity between above-mentioned domain color feature calculation image, and utilize cluster index table to carry out cluster and quick-searching to image.
Fig. 2 is the system architecture schematic diagram that the disclosure realizes image retrieval.
As shown in Figure 2, image characteristic analysis extraction module and image retrieval module can be comprised.
Image characteristic analysis extraction module is responsible for processing the image in image data base, and the domain color realizing image extracts, and characteristics of image batch is deposited in image data base; Index value synthesis cluster index table name corresponding to the domain color extracted claims, and the index of image is saved in cluster index table, sets up cluster index database.The function of signature analysis extraction module realizes as follows: image reading, color of image spatial transformation, domain color feature extraction and preservation index.Detailed step is as follows:
(1) image reading: the title obtaining all images in image data base, forms image name list, and in computed image database, institute comprises the total A of image, and uses 1 ~ A to distinguish image in marker image database.According to mark number, successively the image in image data base is read in signature analysis extraction module, wherein, the image read in uses RGB color space to represent in systems in which.
(2) color of image space transforming: compare RGB color space, HSV can digitized processing color better.It is quantitatively described by tone H, saturation degree S and the color of brightness V tri-components to image, and wherein, H angle [0 °, 360 °] represents, more adequately reflects the understanding mode of human visual system to color, better to image area calibration.Therefore, first carry out the conversion of RGB to HSV image space, then extract the color characteristic of image in HSV space.Image is after completing color space conversion, and each pixel in image all uses tone h, saturation degree s and brightness v to represent, wherein, the span of h be 0 to 360, s span be 0 to 1, v span be 0 to 1.
Fig. 3 is the schematic diagram that the disclosure is changed by RGB to HSV.
As shown in Figure 3, the conversion of brightness v, saturation degree s and tone h can be carried out respectively.Such as, the following quantization algorithm of HSV space at equal intervals can be adopted:
By above algorithm, tone h is divided into 9 parts, saturation degree s and brightness v is divided into 3 parts.Therefore, the color after quantification has 81 dimensions.
(3) domain color feature extraction: after image is transformed into HSV space by RGB, HSV image volume is changed into 81 dimension color characteristics, select front 8 dimensional features of eigenwert descending sort by size, thus realize the extraction to image domain color, and characteristics of image in batches stored in image data base.
The descriptor of image domain color can use formula F={ { C
i, P
i, i=1,2,3 ..., 81, P
i∈ [0,1] } definition, wherein, C
ibe a three-dimensional mass-tone vector (C=9*H+3*S+V) represented with H, S, V, Pi is mass-tone number percent.
(4) index is preserved: 81 index values that in step (3), the subscript of 81 dimension color characteristics is corresponding, corresponding to the 8 dimension domain color extracted, 8 index values are combined into cluster index table name and claim, and the index of image is saved in cluster index table, set up cluster index database, quantized color c={1,2 ... 81}, corresponding index value is C
i={ C
1, C
2..., C
81.
Image retrieval module is responsible for first carrying out image characteristic analysis extraction to image to be retrieved, secondly be combined into cluster index table name according to the domain color index value of this image to claim, image index in search index table, according to Query Result inquiry mass-tone number percent, calculate the similarity between image to be retrieved and the image inquired in image data base, the matching image satisfied condition is returned to user.The function of this module can realize as follows: the extraction of retrieving images, color of image space transforming, signature analysis, search index and computed image similarity.Detailed step is as follows:
(5) retrieving images: obtain image to be retrieved from terminal, read in image retrieval module.The image read in uses RGB color space to represent in systems in which.
(6) color of image space transforming: be transformed into HSV space by rgb space, after color of image space transforming completes, each pixel of image uses tone h, saturation degree s and brightness v represents, wherein, the span of h is 0 to 360, the span of s be 0 to 1, v span be 0 to 1.
(7) signature analysis extracts: can be quantized into 81 dimension color characteristics after image to be retrieved is transformed into HSV space, selects the front 8 dimension color characteristics of eigenwert descending sort by size, thus realizes extraction to image domain color.
(8) search index table: be combined into cluster index table name according to the domain color index value of the image to be retrieved obtained in step (7) and claim, then inquire about the image index in the concordance list of above-mentioned title, quantized color c={1,2,, 81}, corresponding index value is C
i={ C
1, C
2..., C
81.
(9) similarity between computed image: according to the image mass-tone number percent of the search index image data base that step (8) checks out, calculate the similarity between image to be retrieved and the image inquired from image data base, the matching image satisfied condition is returned to user.
Image retrieval procedure based on MPEG-7 dominant color descriptors is as follows:
(9a) Q is used
jrepresent P
inumber percent descending, get a front M color as mass-tone, non-mass-tone will not be considered.
(9b) the response number percent of a normalized M mass-tone
(9c) F
q={ C
qi, P'
qiand F
t={ C
ti, P'
tibe respectively the dominant color descriptors of image Q and image T.
Similarity is:
Wherein, similarity D is more close to 1, and the similarity of two images is higher, otherwise more close to 0, similarity is lower.In different image data bases, number of dominant colors can change, experiment prove, domain color M=8 be enough used for represent based on MPEG-7 characteristics of image and obtain good retrieval effectiveness.
One of ordinary skill in the art will appreciate that, realize the whole of said method embodiment to have been come by the hardware that programmed instruction is relevant with part steps, aforesaid program can be stored in a computing equipment read/write memory medium, this program is when performing, perform and comprise the step of said method embodiment, and aforesaid storage medium can comprise ROM, RAM, magnetic disc and CD etc. various can be program code stored medium.
Fig. 4 is the structural representation of the image retrieving apparatus of a disclosure embodiment.
As shown in Figure 4, the device 40 in this embodiment can comprise image acquisition unit 402, space conversion unit 404, image quantization unit 406, domain color determining unit 408, concordance list title determining unit 410, judging unit 412, mass-tone number percent acquiring unit 414, similarity calculated 416 and result and return unit 418.Wherein,
Image acquisition unit 402, obtains image to be retrieved for user terminal, and image to be retrieved uses rgb space to represent;
Space conversion unit 404, for being transformed into HSV space by image to be retrieved by rgb space;
Image quantization unit 406, for by the image quantization to be retrieved being transformed into HSV space being N dimension color characteristic;
Domain color determining unit 408, for the N of image to be retrieved being tieed up the eigenwert descending sort by size of color characteristic, choosing the front M after sequence and tieing up the domain color of color characteristic as image to be retrieved, wherein, and M<N;
For the index value corresponding according to the domain color of image to be retrieved, concordance list title determining unit 410, determines that the cluster index table name of image to be retrieved claims;
Judging unit 412, claims identical cluster index table for whether existing in the cluster index table set up based on image each in image data base according to the cluster index table name query of the image to be retrieved determined with the cluster index table name of image to be retrieved;
Mass-tone number percent acquiring unit 414, for such as existing, then from claiming to obtain image index in identical cluster index table with the cluster index table name of image to be retrieved, and the image mass-tone number percent corresponding according to the image index inquiry obtained;
Similarity calculated 416, for based on the similarity between image mass-tone percentage calculation image to be retrieved and the image inquired from image data base;
Result returns unit 418, for the image with images match to be retrieved being returned to user terminal according to the similarity between image.
In this embodiment, owing to by the image quantization in HSV space being N dimension, and according to the domain color of eigenwert determination image, the cluster index table name of the domain color determination image of image is utilized to claim, first the image with identical domain color is searched when treating retrieving images and retrieving, but retrieve in the image with identical domain color again, visible, the scope of retrieval is significantly reduced compared with existing image search method.Meanwhile, due to the range of search of the domain color determination image according to image, the accuracy of image retrieval is also further ensured.
Further, this device can also comprise:
Unit set up by concordance list, for setting up cluster index table based on image each in image data base.
Wherein, unit set up by concordance list can comprise image acquisition unit, space conversion unit, image quantization unit, domain color determining unit, concordance list title determining unit and storing sub-units, wherein,
Image acquisition unit, also for each image in reads image data storehouse successively, wherein, each image in image data base uses rgb space to represent;
Space conversion unit, also for each image in image data base is transformed into HSV space by rgb space;
Image quantization unit, also for by each image quantization being transformed into HSV space being N dimension color characteristic;
Domain color determining unit, also for the N of each image being tieed up the eigenwert descending sort by size of color characteristic, choosing the front M after sequence and tieing up the domain color of color characteristic as each image;
Concordance list title determining unit, also for claiming according to the cluster index table name of each image in index value determination image data base corresponding to the domain color of each image;
Storing sub-units, for being saved in the image index value of each image and corresponding mass-tone number percent in corresponding cluster index table.
Further, the tone H in HSV space can be divided into 9 parts equably, saturation degree S can be divided into 3 parts by image quantization unit, and brightness V is divided into 3 parts.
In this instructions, each embodiment all adopts the mode of going forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, and part identical with similar between each embodiment can cross-reference.For device embodiment, due to itself and embodiment of the method basic simlarity, so description is fairly simple, relevant part can see the explanation of embodiment of the method part.
Disclosure above-described embodiment hinge structure tool has the following advantages and effect:
(1) searching algorithm of the present disclosure dimension when extracting the color characteristic of image is 81 dimensions, makes extracted characteristic information more accurate, improves the precision ratio of system.
(2) in HSV model, red corresponding to h=0 °, green corresponding to h=120 °, blue corresponding to h=240 °, the disclosure is also close red according to the both sides [360 °, 20 °] of red (0 °), [100 °, the both sides of green (120 °), 140 °] also close to green, the disclosure is more reasonable to the division in colourity h interval, improves the recall ratio of system.
(3) color characteristic of image is quantized into 81 dimensions by searching algorithm of the present disclosure, and the domain color of retrieving images only accounts for 8 dimensions, and therefore only some image belongs in certain 8 dimension; Only concordance list index being carried out to the image satisfied condition when setting up index, being provided with restriction querying condition simultaneously, only just search index is carried out to the concordance list satisfied condition, very low with total image number degree of correlation of image data base, significantly reduce retrieval time.
(4) system only needs when image retrieval to retrieve the dendrogram picture belonging to retrieving images, ignores the image of other clusters, avoids the invalid retrieval to a large amount of dissimilar image, thus significantly improve validity and the efficiency of retrieval.
Although describe the disclosure with reference to exemplary embodiment, should be understood that the disclosure is not limited to above-mentioned exemplary embodiment.It will be obvious to those skilled in the art that and can revise above-mentioned exemplary embodiment under the condition not deviating from the scope of the present disclosure and spirit.The scope of appended claim should be endowed the widest explanation, to comprise all such amendments and equivalent 26S Proteasome Structure and Function.
Claims (8)
1. an image search method, is characterized in that, comprising:
User terminal obtains image to be retrieved, and described image to be retrieved uses rgb space to represent;
Image to be retrieved is transformed into HSV space by rgb space;
Be that N ties up color characteristic by the image quantization to be retrieved being transformed into HSV space;
The N of image to be retrieved is tieed up the eigenwert descending sort by size of color characteristic, choose the front M after sequence and tie up the domain color of color characteristic as image to be retrieved, wherein, M<N;
The index value corresponding according to the domain color of image to be retrieved determines that the cluster index table name of image to be retrieved claims;
Whether exist in the cluster index table set up based on image each in image data base according to the cluster index table name query of the image to be retrieved determined and claim identical cluster index table with the cluster index table name of image to be retrieved;
As existed, then from claiming to obtain image index in identical cluster index table with the cluster index table name of image to be retrieved;
The image mass-tone number percent corresponding according to the image index inquiry obtained;
Based on the similarity between image mass-tone percentage calculation image to be retrieved and the image inquired from image data base;
According to the similarity between image, the image with images match to be retrieved is returned to user terminal.
2. image search method according to claim 1, is characterized in that, described method also comprises:
Cluster index table is set up based on image each in image data base.
3. image search method according to claim 2, is characterized in that, the described step setting up cluster index table based on image each in image data base comprises:
Each image successively in reads image data storehouse, wherein, each image in image data base uses rgb space to represent;
Each image in image data base is transformed into HSV space by rgb space;
Be that N ties up color characteristic by each image quantization being transformed into HSV space;
The N of each image is tieed up the eigenwert descending sort by size of color characteristic, choose the front M after sequence and tie up the domain color of color characteristic as each image;
Claim according to the cluster index table name of each image in the index value determination image data base that the domain color of each image is corresponding;
The image index value of each image and corresponding mass-tone number percent are saved in corresponding cluster index table.
4. image search method according to claim 3, is characterized in that, the tone H in HSV space is divided into 9 parts equably, saturation degree S is divided into 3 parts, brightness V is divided into 3 parts.
5. an image retrieving apparatus, is characterized in that, comprising:
Image acquisition unit, obtains image to be retrieved for user terminal, and described image to be retrieved uses rgb space to represent;
Space conversion unit, for being transformed into HSV space by image to be retrieved by rgb space;
Image quantization unit, for by the image quantization to be retrieved being transformed into HSV space being N dimension color characteristic;
Domain color determining unit, for the N of image to be retrieved being tieed up the eigenwert descending sort by size of color characteristic, choosing the front M after sequence and tieing up the domain color of color characteristic as image to be retrieved, wherein, and M<N;
For the index value corresponding according to the domain color of image to be retrieved, concordance list title determining unit, determines that the cluster index table name of image to be retrieved claims;
Judging unit, claims identical cluster index table for whether existing in the cluster index table set up based on image each in image data base according to the cluster index table name query of the image to be retrieved determined with the cluster index table name of image to be retrieved;
Mass-tone number percent acquiring unit, for such as existing, then from claiming to obtain image index in identical cluster index table with the cluster index table name of image to be retrieved, and the image mass-tone number percent corresponding according to the image index inquiry obtained;
Similarity calculated, for based on the similarity between image mass-tone percentage calculation image to be retrieved and the image inquired from image data base;
Result returns unit, for the image with images match to be retrieved being returned to user terminal according to the similarity between image.
6. image retrieving apparatus according to claim 5, is characterized in that, described device also comprises:
Unit set up by concordance list, for setting up cluster index table based on image each in image data base.
7. image retrieving apparatus according to claim 6, it is characterized in that, described concordance list is set up unit and is comprised described image acquisition unit, described space conversion unit, described image quantization unit, described domain color determining unit, described concordance list title determining unit and storing sub-units, wherein
Described image acquisition unit, also for each image in reads image data storehouse successively, wherein, each image in image data base uses rgb space to represent;
Described space conversion unit, also for each image in image data base is transformed into HSV space by rgb space;
Described image quantization unit, also for by each image quantization being transformed into HSV space being N dimension color characteristic;
Described domain color determining unit, also for the N of each image being tieed up the eigenwert descending sort by size of color characteristic, choosing the front M after sequence and tieing up the domain color of color characteristic as each image;
Described concordance list title determining unit, also for claiming according to the cluster index table name of each image in index value determination image data base corresponding to the domain color of each image;
Described storing sub-units, for being saved in the image index value of each image and corresponding mass-tone number percent in corresponding cluster index table.
8. image retrieving apparatus according to claim 7, is characterized in that, the tone H in HSV space is divided into 9 parts by image quantization unit equably, saturation degree S is divided into 3 parts, and brightness V is divided into 3 parts.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410175999.3A CN105022752B (en) | 2014-04-29 | 2014-04-29 | Image search method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410175999.3A CN105022752B (en) | 2014-04-29 | 2014-04-29 | Image search method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105022752A true CN105022752A (en) | 2015-11-04 |
| CN105022752B CN105022752B (en) | 2019-04-05 |
Family
ID=54412734
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410175999.3A Active CN105022752B (en) | 2014-04-29 | 2014-04-29 | Image search method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105022752B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105787965A (en) * | 2016-01-26 | 2016-07-20 | 安徽创世科技股份有限公司 | Image searching method based on color features |
| CN105912611A (en) * | 2016-04-05 | 2016-08-31 | 中国科学技术大学 | CNN based quick image search method |
| CN106780634A (en) * | 2016-12-27 | 2017-05-31 | 努比亚技术有限公司 | Picture dominant tone extracting method and device |
| CN107832359A (en) * | 2017-10-24 | 2018-03-23 | 杭州群核信息技术有限公司 | A kind of picture retrieval method and system |
| CN107908630A (en) * | 2017-06-28 | 2018-04-13 | 重庆完美空间科技有限公司 | Material picture color classification retrieving method |
| CN108764352A (en) * | 2018-05-25 | 2018-11-06 | 百度在线网络技术（北京）有限公司 | Duplicate pages content detection algorithm and device |
| CN109308325A (en) * | 2018-08-21 | 2019-02-05 | 董志忠 | Image search method and system |
| CN109558506A (en) * | 2018-11-29 | 2019-04-02 | 青海民族大学 | A kind of image search method based on color convergence vector |
| CN110119460A (en) * | 2019-05-16 | 2019-08-13 | 广东三维家信息科技有限公司 | Image search method, device and electronic equipment |
| CN110502651A (en) * | 2019-08-15 | 2019-11-26 | 深圳市商汤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN110647910A (en) * | 2019-08-12 | 2020-01-03 | 浙江浩腾电子科技股份有限公司 | Image similarity calculation method based on color quantization |
| CN110879849A (en) * | 2019-11-09 | 2020-03-13 | 广东智媒云图科技股份有限公司 | Similarity comparison method and device based on image-to-character conversion |
| CN112131424A (en) * | 2020-09-22 | 2020-12-25 | 深圳市天维大数据技术有限公司 | Distributed image analysis method and system |
| WO2021012521A1 (en) * | 2019-07-19 | 2021-01-28 | 平安科技（深圳）有限公司 | Search-based webpage forensics method and device, readable storage medium and server |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101714257A (en) * | 2009-12-23 | 2010-05-26 | 公安部第三研究所 | Method for main color feature extraction and structuring description of images |
| CN102542014A (en) * | 2011-12-16 | 2012-07-04 | 华中科技大学 | Image searching feedback method based on contents |
| CN102722880A (en) * | 2011-03-29 | 2012-10-10 | 阿里巴巴集团控股有限公司 | Image main color identification method and apparatus thereof, image matching method and server |
| CN103366178A (en) * | 2012-03-30 | 2013-10-23 | 北京百度网讯科技有限公司 | Method and device for carrying out color classification on target image |
-
2014
- 2014-04-29 CN CN201410175999.3A patent/CN105022752B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101714257A (en) * | 2009-12-23 | 2010-05-26 | 公安部第三研究所 | Method for main color feature extraction and structuring description of images |
| CN102722880A (en) * | 2011-03-29 | 2012-10-10 | 阿里巴巴集团控股有限公司 | Image main color identification method and apparatus thereof, image matching method and server |
| CN102542014A (en) * | 2011-12-16 | 2012-07-04 | 华中科技大学 | Image searching feedback method based on contents |
| CN103366178A (en) * | 2012-03-30 | 2013-10-23 | 北京百度网讯科技有限公司 | Method and device for carrying out color classification on target image |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105787965B (en) * | 2016-01-26 | 2018-08-07 | 安徽创世科技股份有限公司 | A kind of image search method based on color characteristic |
| CN105787965A (en) * | 2016-01-26 | 2016-07-20 | 安徽创世科技股份有限公司 | Image searching method based on color features |
| CN105912611A (en) * | 2016-04-05 | 2016-08-31 | 中国科学技术大学 | CNN based quick image search method |
| CN105912611B (en) * | 2016-04-05 | 2019-04-26 | 中国科学技术大学 | A kind of fast image retrieval method based on CNN |
| CN106780634B (en) * | 2016-12-27 | 2019-06-18 | 努比亚技术有限公司 | Picture dominant tone extracting method and device |
| CN106780634A (en) * | 2016-12-27 | 2017-05-31 | 努比亚技术有限公司 | Picture dominant tone extracting method and device |
| CN107908630A (en) * | 2017-06-28 | 2018-04-13 | 重庆完美空间科技有限公司 | Material picture color classification retrieving method |
| CN107832359B (en) * | 2017-10-24 | 2021-06-08 | 杭州群核信息技术有限公司 | Picture retrieval method and system |
| CN107832359A (en) * | 2017-10-24 | 2018-03-23 | 杭州群核信息技术有限公司 | A kind of picture retrieval method and system |
| CN108764352A (en) * | 2018-05-25 | 2018-11-06 | 百度在线网络技术（北京）有限公司 | Duplicate pages content detection algorithm and device |
| CN108764352B (en) * | 2018-05-25 | 2022-09-27 | 百度在线网络技术（北京）有限公司 | Method and device for detecting repeated page content |
| CN109308325A (en) * | 2018-08-21 | 2019-02-05 | 董志忠 | Image search method and system |
| CN109308325B (en) * | 2018-08-21 | 2022-07-01 | 董志忠 | Image searching method and system |
| CN109558506A (en) * | 2018-11-29 | 2019-04-02 | 青海民族大学 | A kind of image search method based on color convergence vector |
| CN110119460A (en) * | 2019-05-16 | 2019-08-13 | 广东三维家信息科技有限公司 | Image search method, device and electronic equipment |
| WO2021012521A1 (en) * | 2019-07-19 | 2021-01-28 | 平安科技（深圳）有限公司 | Search-based webpage forensics method and device, readable storage medium and server |
| CN110647910A (en) * | 2019-08-12 | 2020-01-03 | 浙江浩腾电子科技股份有限公司 | Image similarity calculation method based on color quantization |
| CN110502651B (en) * | 2019-08-15 | 2022-08-02 | 深圳市商汤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN110502651A (en) * | 2019-08-15 | 2019-11-26 | 深圳市商汤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN110879849A (en) * | 2019-11-09 | 2020-03-13 | 广东智媒云图科技股份有限公司 | Similarity comparison method and device based on image-to-character conversion |
| CN112131424A (en) * | 2020-09-22 | 2020-12-25 | 深圳市天维大数据技术有限公司 | Distributed image analysis method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105022752B (en) | 2019-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105022752A (en) | Image retrieval method and apparatus | |
| CN102819582B (en) | Quick searching method for mass images | |
| Singh et al. | Content-based image retrieval using color moment and Gabor texture feature | |
| Afifi et al. | Image retrieval based on content using color feature | |
| Yang et al. | An improved Bag-of-Words framework for remote sensing image retrieval in large-scale image databases | |
| CN102890700A (en) | Method for retrieving similar video clips based on sports competition videos | |
| Singh et al. | Image retrieval based on the combination of color histogram and color moment | |
| Climer et al. | Image database indexing using JPEG coefficients | |
| Brown | Example-based color vehicle retrieval for surveillance | |
| Tolias et al. | Towards large-scale geometry indexing by feature selection | |
| Feng et al. | An efficient indexing method for content-based image retrieval | |
| Song et al. | Content-based image retrieval using new color histogram | |
| Shi et al. | Image retrieval using both color and texture features | |
| Jaswal et al. | Content based image retrieval using color space approaches | |
| Kavitha et al. | Image retrieval based on local histogram and texture features | |
| Fierro-Radilla et al. | An effective visual descriptor based on color and shape features for image retrieval | |
| Huang et al. | A method for object-based color image retrieval | |
| Asif et al. | Composite description based on color vector quantization and visual primary features for CBIR tasks | |
| Rusinol et al. | Perceptual image retrieval by adding color information to the shape context descriptor | |
| Sebai et al. | Improving high resolution satellite images retrieval using color component features | |
| Bouker et al. | Retrieval of images using mean-shift and gaussian mixtures based on weighted color histograms | |
| Lee et al. | Efficient object identification and localization for image retrieval using query-by-region | |
| Gao et al. | Multiple features-based image retrieval | |
| Goswami et al. | RISE: a robust image search engine | |
| CN107870923B (en) | Image retrieval method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |