CN103366178A

CN103366178A - Method and device for carrying out color classification on target image

Info

Publication number: CN103366178A
Application number: CN2012100913538A
Authority: CN
Inventors: 文林福
Original assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Current assignee: Beijing Baidu Netcom Science and Technology Co Ltd
Priority date: 2012-03-30
Filing date: 2012-03-30
Publication date: 2013-10-23
Anticipated expiration: 2032-03-30
Also published as: CN103366178B

Abstract

The invention aims to provide a method and a device for carrying out color classification on a target image. Specifically, the method comprises the steps of dividing an HSV color space into a plurality of color subspaces, determining a mapping relation between partial or all pixels in the target image and the plurality of color subspaces according to an HSV assignment of the target image in the HSV color space so as to determine color classifications of the partial or all pixels, and determining the color classification of the target image according to the color classifications of the partial or all pixels. The method and the device provided by the invention solve a problem that an existing method can not apply to various different application scenarios, improve the accuracy of a color classification method, and further improve the efficiency of the color classification method of the target image and the usage experience of users.

Description

A kind of method and apparatus for target image being carried out color classification

Technical field

The present invention relates to technical field of image processing, relate in particular to a kind of technology of image being carried out color classification based on the hsv color space.

Background technology

Along with the develop rapidly of electronic technology and digital processing technology, the raising of the image electronic equipment popularity rates such as digital camera, scanner, picture material increases just with surprising rapidity.How effectively retrieving required image from the view data of magnanimity is a very crucial problem.For example the user wants to retrieve a photo similar to Beethovan's appearance, and general text search engine is almost felt simply helpless for graph image usually.Therefore, effective retrieval of image become the key issue that we obtain image information.

Present image retrieval (Image Retrieval) adopts two kinds of methods usually: one is based on the image retrieval of text index, and two are based on the image retrieval of content.System based on the image retrieval of text index, it all is the semanteme that comes first Description Image with text, then use ripe text search algorithm on these image text mark bases, such as Page-Rank method, probabilistic method, location method, method of abstracting, classification or clustering method, part-of-speech tagging method etc., be the image of user search expection.But the increase along with the image data base scale, more than several hundred million, the problem that exists based on the image retrieval of text index has just highlighted, at first manually image being marked too wastes time and energy, secondly artificial mark has subjectivity and uncertainty, as for same width of cloth image, the mark that different people provides may be fully different, and this is to respond exactly user's inquiry very difficult.CBIR does not need user's participation, and utilizes the feature of image self, retrieves such as features such as color, texture, shapes, has stronger objectivity.Usually, all Characteristic of Images in can the abstract image storehouse, the process of user search generally provides a sample image, system extracts the individual features of this sample image, then compare with all features of the object that is retrieved in the database, and will return to the user with the image of sample feature similarity.

Color characteristic is based on the visual signature that is most widely used in the image retrieval of content, main cause be color often with image in the object or the scene that comprise very relevant.In addition, compare with other visual signature, color characteristic is less to the dependence at the size of image itself, direction, visual angle, thereby has higher robustness.Color characteristic method for expressing at present commonly used has the method for expressing of the color characteristics such as color histogram, color moment, color set, color convergence vector and color correlogram.At present the statistic that target image is converted to the above-mentioned color characteristic of employing by image processing techniques is mainly adopted in the color classification of target image, as color space being divided between several little chromatic zoneses, then drop on the color rarity that pixel quantity in each minizone can obtain target image by calculating color.Present method for sorting colors can't be taken into account performance and efficient requirement, and is many such as the number of color minizone, just stronger to the resolution characteristic of color, but can increase computation burden, is unfavorable for being applied in Large Graph as setting up index in the library searching; The number of color minizone is few, although computing cost is less, the resolution characteristic of color is lower, is unfavorable for being applied in the application that can't stand the associated picture mistakes and omissions.A core key technology taking into account the method for sorting colors of performance and efficient is to solve color space subspace partition problem, to obtain a plurality of color sub-spaces, then determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel.Yet present target image method for sorting colors does not solve color space subspace partition problem, and then causes present method for sorting colors can not be applicable to various application scenarios; Present method for sorting colors is not distinguished the band of position of pixel in the target image yet, each pixel in the target image shared weight in color classification is the same, and then cause present method for sorting colors not carry out color classification to target image exactly, be left black right white such as piece image, another width of cloth image is right black left white, wherein black-white colors respectively accounts for 50%, and the histogram of described two width of cloth colors is the same, but this two width of cloth image difference is very obvious.Obviously, the unresolved color space of existing method subspace partition problem, and then cause existing method can not be applicable to various application scenarios; Existing method is not distinguished the band of position of pixel in the target image yet, and then cause the accuracy of method for sorting colors not high yet, has reduced the accuracy that target image carries out color classification, has reduced simultaneously user's experience.

Therefore, content for above-mentioned two aspects, how can make the method for target image color classification be applicable to various different application scenes, the accuracy that can improve again method for sorting colors becomes the problem that those skilled in the art need solution badly, yet current not yet have for this solution of problem scheme.

Summary of the invention

The purpose of this invention is to provide a kind of method and apparatus for target image being carried out color classification.

According to another aspect of the present invention, provide a kind of by computer implemented method for target image being carried out color classification, the method may further comprise the steps:

X carries out the subspace to the hsv color space and divides processing, to obtain a plurality of color sub-spaces, wherein, the corresponding a kind of color of each color sub-spaces;

Wherein, the method also comprises:

A obtains the target image of intending carrying out color classification;

B is according to the HSV assignment of described target image under described hsv color space, determines in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel;

C determines the color classification of described target image according to the color classification of described part or all of pixel.

According to an aspect of the present invention, also provide a kind of sorting device for target image being carried out color classification, this equipment comprises:

The spatial division device is used for that the subspace is carried out in the hsv color space and divides processing, to obtain a plurality of color sub-spaces, wherein, the corresponding a kind of color of each color sub-spaces;

Wherein, this equipment also comprises:

Image acquiring device is used for obtaining the target image of intending carrying out color classification;

The spatial mappings device, be used for according to the HSV assignment of described target image under described hsv color space, determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel;

Color is determined device, is used for the color classification according to described part or all of pixel, determines the color classification of described target image.

In accordance with a further aspect of the present invention, also provide a kind of browser for target image being carried out color classification.

In accordance with a further aspect of the present invention, also provide a kind of search engine for target image being carried out color classification.

Compared with prior art, the present invention is by becoming a plurality of color sub-spaces with the hsv color spatial division, support that wherein by the method for machine learning color space being carried out the subspace divides, according to the HSV assignment of described target image under described hsv color space, determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, to determine the color classification of described part or all of pixel, color classification according to described part or all of pixel, determine the color classification of described target image, wherein support in conjunction with the band of position of described part or all of pixel in described target image, the color classification of described target image is determined in weighting, solve the problem that existing method can not be applicable to various different application scenes, improve the accuracy of method for sorting colors, further improve the method for sorting colors efficient of target image and user's experience.

Description of drawings

By reading the detailed description that non-limiting example is done of doing with reference to the following drawings, it is more obvious that other features, objects and advantages of the present invention will become:

Fig. 1 illustrates the sorting device schematic diagram that is used for target image is carried out color classification according to one aspect of the invention;

Fig. 2 illustrates the sorting device schematic diagram that is used for target image is carried out color classification in accordance with a preferred embodiment of the present invention;

Fig. 3 illustrate according to a further aspect of the present invention by computer implemented method flow diagram for target image being carried out color classification;

Fig. 4 illustrate in accordance with a preferred embodiment of the present invention by computer implemented method flow diagram for target image being carried out color classification.

Same or analogous Reference numeral represents same or analogous parts in the accompanying drawing.

Embodiment

Below in conjunction with accompanying drawing the present invention is described in further detail.

Fig. 1 illustrates sorting device 1 schematic diagram that is used for target image is carried out color classification according to one aspect of the invention.At this, sorting device 1 includes but not limited to individual host, minicomputer, large scale computer, a plurality of main frame collection, network host, single network server, a plurality of webserver collection, and the cloud of a plurality of server formation.At this, cloud is by consisting of based on a large amount of computing machines of cloud computing (Cloud Computing) or the webserver, and wherein, cloud computing is a kind of of Distributed Calculation, a super virtual machine that is comprised of the loosely-coupled computing machine collection of a group.

As shown in Figure 1, sorting device 1 comprises that spatial division device 101, image acquiring device 102, spatial mappings device 103, color determine device 104.

Particularly, 101 pairs of hsv color spaces of spatial division device are carried out the subspace and are divided processing, to obtain a plurality of color sub-spaces, wherein, the corresponding a kind of color of each color sub-spaces.

Wherein, described HSV (Hue, Saturation, Value) color space is the three basic attribute according to color: form and aspect (Hue), saturation degree (Saturation) and lightness (Value) are determined a kind of method of color.The model in hsv color space is corresponding to a conical subset in the cylindrical-coordinate system, the suitable form and aspect of the rounded bottom surface of cone, and saturation degree then increases to the edge from the center of circle, lightness is then successively decreased to the vertex of a cone from the base, H, S, the span of V is respectively 0≤V≤1,0≤S≤1,0≤H＜360.The end face of circular cone is corresponding to V=1, and representative color is brighter, and color H is by given around the rotation angle of V axle.Red corresponding to 0 ° of angle, green corresponding to 120 ° of angles, blue corresponding to 240 ° of angles.In the hsv color model, each color and its complementary color differ 180 °.Saturation degree S value from 0 to 1, so the radius of circular cone end face is 1, place, the summit of circular cone, V=0, represent black, the end face center S=0 of circular cone, V=1, represent white, represent gradually dark grey of brightness from the end face center of circular cone to the summit of circular cone, namely have the grey of different gray scales.Other color spaces commonly used also have the RGB color space, HSI color space, HSL color space, HSB color space, YUV color space, Lab color space, XYZ color space, Ycc color space, CMYK color space etc.With respect to other color spaces, the hsv color space meets visual color continuity closer to the subjective understanding of people to color, and color distribution is continuous basically, is convenient to the division of subspace.

For example, in the hsv color space, the spatial division device 101 usefulness space plane vertical with the V axle is divided equally into 10 five equilibriums with the V axle, being about to described hsv color spatial division and becoming 10 sub spaces, is 0-360 such as the H value, and the S value is 0-1, the V value is that the color space of 0-0.1 is subspace, a hsv color space, this subspace is labeled as HSV[0-360,0-1,0-0.1].Described color sub-spaces comprises every kind of color corresponding spatial dimension in each color space subspace, for example red at color space subspace HSV[0-360,0-1,0-0.1] in curve be f1 (H, S) with f2 (H, S), drop on curve f1 (H, S) with f2 (H, S) between scope in pixel namely be labeled as redness.

Those skilled in the art can understand and above-mentionedly become 10 the method for subspace only as for example take the hsv color spatial division; other are existing or may occur that from now on the subspace is carried out in the hsv color space and divide process; with the method that obtains a plurality of color sub-spaces as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Image acquiring device 102 obtains the target image of intending carrying out color classification.

Wherein, the form of described target image includes but not limited to: bmp, jpg, tiff, gif, pcx, tga, exif, fpx, svg, psd, cdr, pcd, dxf, ufo, eps, ai, raw.

Wherein, described target image comprise be stored in local and or network computer or memory device in, wherein network computer or memory device comprise cloud computing.

Wherein, described target image comprise by wired and or wireless mode from described computing machine or memory device, obtain.

For example image acquiring device 102 passes through the internet, use the HTTPs agreement, the inquiry API that provides by database from the network picture database obtains the target image that 1000 plans are carried out color classification, such as " Garfield " image, its storage format is the picture of jpg.Wherein the network picture database comprises and being stored on the special-purpose server, and it is first-class to be stored in cloud computing platform.

Those skilled in the art can understand the above-mentioned internet that passes through; use the HTTPs agreement; the method of obtaining the target image of intending carrying out color classification from the network picture database is only for for example; other are existing or may occur from now on be used to the method for obtaining the target image of intending carrying out color classification as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Spatial mappings device 103, according to the HSV assignment of described target image under described hsv color space, determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel.

Wherein said HSV assignment refers to the corresponding HSV value of each pixel in the target image, and the HSV that for example the pixel x in the target image is corresponding is H=0, S=0.5, and V=0.05 is designated as HSV[0, and 0.5,0.05].

Wherein said mapping relations comprise the HSV value according to each pixel in the target image, and the division of color space subspace, determine the subspace that this pixel is affiliated, then shine upon with the color sub-spaces of this subspace.

For example, the HSV[0 of pixel r among the target image e, 0.5,0.05], the hsv color space uses the space plane vertical with the V axle that the V axle is divided equally into 10 five equilibriums, be about to the sub spaces that described hsv color spatial division becomes 10 sub spaces, the subspace is HSV[0-360 under one of them, 0-1,0-0.1], spatial mappings device 103 determines that described pixel r belongs to subspace HSV[0-360,0-1,0-0.1], spatial mappings device 103 shines upon the HSV value of pixel r and the color sub-spaces of this subspace, determines that this pixel r drops between red curve f1 and the f2, and then determines that this pixel r belongs to red, spatial mappings device 103 can adopt the mode of batch processing or the mode of distributed treatment that part or all of pixel in the target image is shone upon, to determine the color classification of described part or all of pixel.

Those skilled in the art can understand the HSV value of above-mentioned pixel r in target image e and determine that the method for color classification is only as giving an example; other are existing or may occur from now on according to the HSV assignment of described target image under described hsv color space; determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces; with the method for the color classification of determining described part or all of pixel as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Color is determined device 104, according to the color classification of described part or all of pixel, determines the color classification of described target image.

For example, determine the color classification of described target image according to the statistical information of color classification, the statistical information of the pixel color of target image classification is: redness accounts for 60%, scarletly accounts for 15%, orange redly account for 10%, bright red accounts for 2%, and gold accounts for 2%, and green accounts for 3%, deep sky bluely account for 3%, color is determined the statistical information that device 104 is classified according to the pixel color of target image, analyzes red relevant colors and accounts for 92%, determines that the color classification of described target image is for red.

Those skilled in the art can understand above-mentioned method take the color classification of determining described target image by the statistical information of color classification only as for example; other are existing or color classification according to described part or all of pixel may occur from now on; determine that the method for color classification of described target image is as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Preferably, be continuous firing between each device of sorting device 1, particularly, spatial division device 11 continues that the subspace is carried out in the hsv color space and divides processing, to obtain a plurality of color sub-spaces; Image acquiring device 12 continues to obtain the target image of intending carrying out color classification; Spatial mappings device 13, continue according to the HSV assignment of described target image under described hsv color space, determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel; Color is determined device 14, continues the color classification according to described part or all of pixel, determines the color classification of described target image.Constant work between above-mentioned each device, at this, it will be understood by those skilled in the art that " continuing " refers to that above-mentioned each device requires to carry out the hsv color space according to the mode of operation of setting or adjust in real time respectively and carries out the subspace and divide and process, obtain the target image of intending carrying out color classification, the color classification of determining described part or all of pixel, and the color classification of determining described target image.

Preferably, the spatial division device 101:

-by machine learning, the subspace is carried out in the hsv color space divide processing, to obtain described a plurality of color sub-spaces, wherein, the corresponding a kind of color of each color sub-spaces.

Wherein, described machine learning comprises by computer simulation or realizes human learning behavior, and self-correction, automatically improves.

For example, spatial division device 101 carries out the subspace to the hsv color space and divides processing by machine learning, and to obtain the red color subspace, the step of machine learning is as follows:

Step 1: the positive negative sample of having collected first red color, positive sample is thought red sample for people's subjectivity and is joined positive sample set, negative sample thinks that for people's subjectivity the sample that is not red adds the negative sample set, wherein color card is collected and can be passed through repeatedly the iteration realization, and this sample claims marker color sample;

Step 2: in the hsv color space, use the space plane vertical with the V axle that the V axle is divided equally into 10 five equilibriums, be about to described hsv color spatial division and become 10 sub spaces;

Step 3: in described every sub spaces, set first initial red curve, such as the red curve formula be: f1 (H, S)=H+S^a1*b1+c1, f2 (H, S)=H+S^a2*b2+c2, wherein a, b, c is parameter, a, b, the span of c is respectively: a (1/10050), b (360360), c (360360), the HSV value of target image pixel drops on the redness that is labeled as between two curves;

Wherein initial parameter of curve is set to f1 (H, S)=H+S^0*0+0=0, f2 (H, S)=H+S^0*0+30=0;

Step 4: use initial formula that marker color sample is classified, obtain the correct and wrong ratio of classification;

Step 5: according to a, b, the span of c, constantly update formula f1 and f2 according to step-length, step-length of every renewal is classified to marker color sample, obtaining the ratio of a correct and classification error, is 5, a as c being got step-length, b is constant, obtain once new formula: f (H, S)=H+5=0 and f (H, S)=H+35=0, then use more new formula that marker color sample is classified, obtain the correct and wrong ratio of classification;

Step 6: after all possible step size combination all tested, obtain one group of optimum parameter, i.e. a group of classification error rate minimum;

Step 7: use one group of optimum parameter, unmarked color card is classified, the color card of misclassification is joined accordingly in the marker color sample, for example red sample is assigned in the negative sample set, this sample just joins in the positive sample set of mark so, color card if not redness is assigned in the positive sample set, then joins in the negative sample set;

Step 8: according to the sample set that upgrades, continuing execution in step 1, carry out iteration, until classification accuracy reaches predetermined threshold, stop iteration, is 95% such as the accuracy rate threshold value;

Step 9: each color is carried out respectively the training of above-mentioned steps 1 to 8.

Those skilled in the art can understand above-mentioned with by machine learning; the subspace is carried out in the hsv color space divide processing; take the method that obtains the red color subspace only for giving an example; other are existing or may occur by machine learning from now on; the subspace is carried out in the hsv color space divide processing;, also should be included in the protection domain of the present invention, and be contained in this with way of reference as applicable to the present invention with the method that obtains the red color subspace.

More preferably, the spatial division device 101:

-according to the color classification number that presets, by machine learning, the subspace is carried out in the hsv color space divide processing, to obtain the described a plurality of color sub-spaces corresponding with described color classification number, wherein, the corresponding a kind of color of each color sub-spaces.

Wherein, the described color classification number that presets comprises that the system that uses the method or user are according to using needs or demand, the color classification number that sets in advance.

For example the user is only interested in scarlet color, and presetting the color classification number is 1, and spatial division device 101 is carried out above-mentioned steps 1 to 8 according to the color classification number, to obtain the corresponding color sub-spaces of scarlet color.

It is 1 by presetting the color classification number that those skilled in the art can understand above-mentioned; by machine learning; take the method that obtains the described a plurality of color sub-spaces corresponding with described color classification number only as giving an example; other are existing or may occur from now on according to the color classification number that presets; by machine learning; the subspace is carried out in the hsv color space divide processing; with the method that obtains the described a plurality of color sub-spaces corresponding with described color classification number as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference

Preferably, color is determined device 104:

-according to color classification and the respective weights thereof of described part or all of pixel, the color classification of described target image is determined in weighting.

For example, color is determined device 104 according to the color classification of each pixel, and the weight of each pixel, and the color classification of described target image is determined in weighting.Perhaps, for example, determine the color classification of described target image by the statistical information weighting of color classification, wherein red weights are 0.1, golden weights are 0.4, green weights are 0.4, blue weights are 0.1, the statistical information of the pixel color classification of target image is: redness accounts for 50%, and gold accounts for 20%, and green accounts for 20%, blueness accounts for 10%, color is determined device 104 according to statistical information and the weights of the pixel color classification of target image, and the statistical information of the pixel of the target image after the calculating weighting is: redness accounts for 22.7%, and gold accounts for 36.3%, green accounts for 36.3%, blueness accounts for 4.5%, and color is determined device 104 according to the pixels statistics information of the target image after the weighting, and the color classification of determining described target image is gold+green.

Those skilled in the art can understand above-mentioned method take the color classification of determining described target image by the statistical information weighting of color classification only as for example; other are existing or color classification and respective weights thereof according to described part or all of pixel may occur from now on; weighting determines that the method for color classification of described target image is as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

More preferably, color is determined device 104:

-according to the color classification of described part or all of pixel, and in conjunction with the band of position of described part or all of pixel in described target image, the color classification of described target image is determined in weighting.

For example, by the color classification that described target image is determined in statistical information and the band of position weighting of pixel in described target image of color classification, a width of cloth target image is square, and its picture material is an one-by-one inch photograph, character image mainly concentrates on center section, use the straight line perpendicular to the base that the target image draw is divided into 3 zones, be respectively left, in, right, wherein yellow weights in left, middle and right are respectively 0.1,0.8,0.1 black is on a left side, in, right weights are respectively 0.1,0.8,0.1, blue on a left side, in, right weights are respectively 0.1,0.8,0.1, the statistical information of the pixel color classification of target image is: yellow accounts for 30%, black accounts for 30%, and blueness accounts for 40%, and is wherein yellow on a left side, in, part on the right side and do not account for 10%, 80%, 10%, black is on a left side, in, part on the right side and do not account for 10%, 80%, 10%, blueness accounts for respectively 40% in left, middle and right, 20%, 40%, color is determined device 104 according to the statistical information of the pixel color classification of target image and the weights of present position, and the statistical information of the target image pixel after the calculating weighting is: yellow accounts for 40.2%, black accounts for 40.2%, blueness accounts for 19.6%, and color is determined device 104 according to the pixels statistics information of the target image after the weighting, and the color classification of determining described target image is yellow+black.

Those skilled in the art can understand above-mentioned statistical information by color classification and the band of position weighting of pixel in described target image determines that the method for the color classification of described target image only is for example; other are existing or color classification according to described part or all of pixel may occur from now on; and in conjunction with the band of position of described part or all of pixel in described target image; weighting determines that the method for color classification of described target image is as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Preferably, sorting device 1 also comprises pretreatment unit 111 (not shown), and 111 pairs of described target images of pretreatment unit carry out pre-service, to obtain pretreated described target image.

Wherein, the pretreatment operation in the described pretreatment unit 111 comprise following at least each:

-described target image is compressed processing;

-described target image is carried out the color space mapping process, to obtain the HSV assignment of described target image under the hsv color space;

-described target image is intercepted processing.

Wherein, before described pre-service is included in the color classification of determining image, this image is carried out pre-service, such as compression, intercepting or color space conversion etc., according to different application scenarioss, target image is carried out certain conversion, and for example the target image color space is the RGB color space, need to be transformed into the HSV image space.

Wherein, described compression processing comprises to be compressed in the situation of not losing too much visual information target image, lossless compression method such as image has the Shannon-Fano coding, the Huffman coding, the distance of swimming (Run-length) coding, LZW (Lempel-Ziv-Welch) coding and arithmetic coding etc., the lossy compression method method of image have Karhunen-Loeve transformation coding and DCT coding etc.Process to reduce the workload of color of image classification by compression, improve color classification efficient.

Wherein, described color space mapping is processed and is comprised that the color space conversion with target image becomes the hsv color space, as the RGB color space conversion is become the hsv color space.

Wherein, subregion or the gold visual zone that comprises by the intercepting target image processed in described intercepting, take the threefold division of image as example, threefold division is vertical with piece image, level is divided into three parts, be that piece image is divided into 9 equal-sized parts in equal size, the joint position of straight line, position namely, be used for the important vision key element of layout, this zone claims the gold visual zone usually, the golden section ratio of image is similar with threefold division, the joint position of the straight line of golden section ratio than the intersection point of threefold division more near picture centre.

For example, become the color space mapping in hsv color space to process the RGB color space conversion of target image, pixel s in the objective definition image is (R, G, B) in the assignment of rgb space, wherein distinguish R, G, B represent respectively red, green and blue assignment, the span of RGB has been done normalization, it is the real number between 0 to 1, if max is R, G, the maximum among the B, min is R, G, the reckling among the B, pretreatment unit 111, each pixel of described target image is carried out the color space mapping process, the RGB color space is as follows to the mapping treatment method in hsv color space:

max＝max(R，G，B)

min＝min(R，G，B)

ifR＝max，H＝(G-B)/(max-min)

ifG＝max，H＝2+(B-R)/(max-min)

ifB＝max，H＝4+(R-G)/(max-min)

H＝H*60

ifH＜0，H＝H+360

V＝max(R，G，B)

S＝(max-min)/max

More preferably, the spatial mappings device 103:

-according to the HSV assignment of described pretreated target image under described hsv color space, determine in the described pretreated target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel.

For example, pretreatment unit 111 converts the RGB color space of target image e to the hsv color space through color space mapping processing, the HSV[0 of pixel r after wherein processing, 0.5,0.05], the hsv color space uses the space plane vertical with the V axle that the V axle is divided equally into 10 five equilibriums, be about to the sub spaces that described hsv color spatial division becomes 10 sub spaces, the subspace is HSV[0-360 under one of them, 0-1,0-0.1], spatial mappings device 103 determines that described pixel r belongs to subspace HSV[0-360,0-1,0-0.1], spatial mappings device 103 shines upon the HSV value of pixel r and the color space of this subspace, determine that this pixel r drops between red curve f1 and the f2, and then determine that this pixel r belongs to red, spatial mappings device 103 can adopt the mode of batch processing or the mode of distributed treatment that part or all of pixel in the target image is shone upon, to determine the color classification of described part or all of pixel.

Those skilled in the art can understand above-mentioned by with the RGB color space of target image e through color space mapping process convert to the hsv color space take the method for the color classification of determining described part or all of pixel only for giving an example; other are existing or may occur from now on according to the HSV assignment of described pretreated target image under described hsv color space; determine in the described pretreated target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces; with the method for the color classification of determining described part or all of pixel as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Fig. 2 illustrates sorting device 1 schematic diagram that is used for target image is carried out color classification in accordance with a preferred embodiment of the present invention, and wherein, sorting device 1 also comprises image providing device 205.Referring to Fig. 2 the preferred embodiment is described: particularly, 201 pairs of hsv color spaces of spatial division device are carried out the subspace and are divided processing, to obtain a plurality of color sub-spaces, wherein, the corresponding a kind of color of each color sub-spaces; Image acquiring device 202 obtains with the user by the corresponding described target image of the search sequence of subscriber equipment input; Spatial mappings device 203 is according to the HSV assignment of described target image under described hsv color space, determines in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel; Color determines that device 204 according to the color classification of described part or all of pixel, determines the color classification of described target image; Image providing device 205 offers described subscriber equipment with the color classification of described target image and described target image.Wherein, spatial division device 201, image acquiring device 202, spatial mappings device 203 and color determine that device 204 is same or similar with corresponding intrument shown in Figure 1 respectively, so locate to repeat no more, and mode by reference is contained in this.Wherein, described subscriber equipment comprises PDA, mobile phone terminal, computing machine, has the digital terminal of network savvy etc.

Particularly, image acquiring device 202 obtains with the user by the corresponding described target image of the search sequence of subscriber equipment input.For example, the user uses mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", image acquiring device 202 uses the HTTPs agreement, call the API that the network picture database provides, obtain the target image relevant with " blue Garfield ".

Those skilled in the art can understand above-mentioned by obtaining with the method for user by the corresponding described target image of the search sequence of mobile phone terminal input only for for example; other are existing or may occur from now on obtaining with the method for user by the corresponding described target image of the search sequence of subscriber equipment input as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Image providing device 205 offers described subscriber equipment with the color classification of described target image and described target image.For example, the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", image acquiring device 202 uses the HTTPs agreement, calls the inquiry API that the network picture database provides, obtain 1000 target images relevant with " blue Garfield ", its storage format is the picture of jpg; Spatial mappings device 203 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; Color determines that device 204 according to the color classification of described whole pixels, determines the color classification of described target image; Image providing device 205 offers described subscriber equipment with the color classification of described target image and described target image, such as the form of color classification by form, and the mode of figure, target image is by the form of file, the form of webpage etc.

Those skilled in the art can understand the above-mentioned method that the color classification of described target image and described target image is offered described subscriber equipment by mobile phone terminal only for for example; other are existing or method that color classification with described target image and described target image offers described subscriber equipment may occur from now on as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Preferably, image providing device 205 according to the color classification of described target image, offers described subscriber equipment with described target image.

For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", image acquiring device 102, use the HTTPs agreement, call the API that the network picture database provides, obtain the image of 1000 " Garfields ", its storage format is the picture of jpg; Spatial mappings device 203 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; Color determines that device 204 according to the color classification of described whole pixels, determines the color classification of described target image; Image providing device 205 offers described subscriber equipment according to the color classification of described target image with described target image, generates different files such as target image according to color classification, and target image presents at the diverse location of webpage according to color classification etc.

Those skilled in the art can understand above-mentioned by the color classification of mobile phone terminal according to described target image; described target image is offered the method for described subscriber equipment only for giving an example; other are existing or color classification according to described target image may occur from now on; described target image is offered described user equipment method as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Preferably, sorting device 1 also comprises Requirement Acquisition device 206 and image matching apparatus 207, and wherein, Requirement Acquisition device 206 obtains the corresponding described user of described search sequence to the color demand information of described target image by semantic analysis; Image matching apparatus 207 is according to color classification and the described color demand information of described target image, the matching image of selecting described color classification and described color demand information to be complementary from described target image; Image providing device 205 offers described subscriber equipment with described matching image.

Particularly, Requirement Acquisition device 206 obtains the corresponding described user of described search sequence to the color demand information of described target image by semantic analysis.Wherein, semantic analysis comprises that the method for using statistical computation analyzes a large amount of text sets, thereby extract semantic structure potential between word and the word, and with this potential semantic structure, obtain the corresponding described user of described search sequence to the color demand information of described target image.

For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", Requirement Acquisition device 206 obtain user's query demand by semantic analysis and are " Garfield ", and color is " blueness ".

Those skilled in the art can understand above-mentioned by mobile phone terminal by semantic analysis; obtain the corresponding described user of described search sequence to the method for the color demand information of described target image only for for example; other are existing or may occur by semantic analysis from now on; obtain the corresponding described user of described search sequence to the method for the color demand information of described target image as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Image matching apparatus 207 is according to color classification and the described color demand information of described target image, the matching image of selecting described color classification and described color demand information to be complementary from described target image.For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", it is Garfield that Requirement Acquisition device 206 obtains user's query demand by semantic analysis, and color is blue, image acquiring device 102, use the HTTPs agreement, call the API that the network picture database provides, obtain the image of 1000 " Garfields ", its storage format is the picture of jpg; Spatial mappings device 203 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; Color determines that device 204 according to the color classification of described whole pixels, determines the color classification of described target image; Image matching apparatus 207 is selected described color classification and the blue Garfield image that is complementary according to color classification and the described color demand information of described target image from described target image.

Those skilled in the art can understand above-mentioned by color classification and the described color demand information of mobile phone terminal according to described target image; the method of the matching image that the described color classification of selection and described color demand information are complementary from described target image is only for giving an example; other are existing or color classification and described color demand information according to described target image may occur from now on; the method of selecting the matching image that described color classification and described color demand information be complementary from described target image is as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Image providing device 205 offers described subscriber equipment with described matching image.

For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", it is Garfield that Requirement Acquisition device 206 obtains user's query demand by semantic analysis, and color is blue, image acquiring device 102, use the HTTPs agreement, call the API that the network picture database provides, obtain the image of 1000 " Garfields ", its storage format is the picture of jpg; Spatial mappings device 203 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; Color determines that device 204 according to the color classification of described whole pixels, determines the color classification of described target image; Image matching apparatus 207 is according to color classification and the described color demand information of described target image, selecting described color classification from described target image is the Garfield image that is complementary with blueness, and image providing device 205 offers described subscriber equipment with described matching image.

It only is for example by mobile phone terminal with the method that described matching image offers described subscriber equipment that those skilled in the art can understand above-mentioned; other existing or methods that may occur from now on described matching image is offered described subscriber equipment are as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Preferably, above-mentioned equipment 1 for target image being carried out color classification can combine with existing browser, consist of a kind of new browser, existing browser can be the IE browser such as Microsoft company, the netscape browser of Netscape company etc.

More preferably, above-mentioned equipment 1 for target image being carried out color classification can also combine with existing search engine, consists of a kind of new search engine.

Fig. 3 illustrate according to a further aspect of the present invention by computer implemented method flow diagram for target image being carried out color classification.At this, realize that the sorting device 1 of the method includes but not limited to individual host, minicomputer, large scale computer, a plurality of main frame collection, network host, single network server, a plurality of webserver collection, and the cloud of a plurality of server formation.At this, cloud is by consisting of based on a large amount of computing machines of cloud computing (Cloud Computing) or the webserver, and wherein, cloud computing is a kind of of Distributed Calculation, a super virtual machine that is comprised of the loosely-coupled computing machine collection of a group.

Particularly, in step S301,1 pair of hsv color space of sorting device is carried out the subspace and is divided processing, to obtain a plurality of color sub-spaces, wherein, the corresponding a kind of color of each color sub-spaces.

For example, in the hsv color space, in step S301, the sorting device 1 usefulness space plane vertical with the V axle is divided equally into 10 five equilibriums with the V axle, is about to described hsv color spatial division and becomes 10 sub spaces, be 0-360 such as the H value, the S value is 0-1, and the V value is that the color space of 0-0.1 is subspace, a hsv color space, and this subspace is labeled as HSV[0-360,0-1,0-0.1].Described color sub-spaces comprises every kind of color corresponding spatial dimension in each color space subspace, for example red at color space subspace HSV[0-360,0-1,0-0.1] in curve be f1 (H, S) with f2 (H, S), drop on curve f1 (H, S) with f2 (H, S) between scope in pixel namely be labeled as redness.

In step S302, sorting device 1 obtains the target image of intending carrying out color classification.

For example in step S302, sorting device 1 uses the HTTPs agreement by the internet, the inquiry API that from the network picture database, provides by database, obtain 1000 plans and carry out the target image of color classification, such as " Garfield " image, its storage format is the picture of jpg.Wherein the network picture database comprises and being stored on the special-purpose server, and it is first-class to be stored in cloud computing platform.

In step S303, sorting device 1 is according to the HSV assignment of described target image under described hsv color space, determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel.

For example, the HSV[0 of pixel r among the target image e, 0.5,0.05], the hsv color space uses the space plane vertical with the V axle that the V axle is divided equally into 10 five equilibriums, be about to the sub spaces that described hsv color spatial division becomes 10 sub spaces, the subspace is HSV[0-360 under one of them, 0-1,0-0.1], in step S303, sorting device 1 determines that described pixel r belongs to subspace HSV[0-360,0-1,0-0.1], sorting device 1 shines upon the HSV value of pixel r and the color sub-spaces of this subspace, determine that this pixel r drops between red curve f1 and the f2, and then determine that this pixel r belongs to red, in step S301, sorting device 1 can adopt the mode of batch processing or the mode of distributed treatment that part or all of pixel in the target image is shone upon, to determine the color classification of described part or all of pixel.

In step S304, sorting device 1 is determined the color classification of described target image according to the color classification of described part or all of pixel.

For example, determine the color classification of described target image according to the statistical information of color classification, the statistical information of the pixel color classification of target image is: redness accounts for 60%, scarletly account for 15%, orange redly account for 10%, bright red accounts for 2%, and gold accounts for 2%, green accounts for 3%, deep sky bluely account for 3%, in step S304, sorting device 1 is according to the statistical information of the pixel color classification of target image, analyze red relevant colors and account for 92%, determine that the color classification of described target image is for red

Preferably, be continuous firing between each step, particularly, in step S301, sorting device 1 continues that the subspace is carried out in the hsv color space and divides processing, to obtain a plurality of color sub-spaces; In step S302, sorting device 1 continues to obtain the target image of intending carrying out color classification; In step S303, sorting device 1 continues according to the HSV assignment of described target image under described hsv color space, determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel; In step S304, sorting device 1 continues the color classification according to described part or all of pixel, determines the color classification of described target image.Constant work between the above steps, at this, it will be understood by those skilled in the art that " continuing " refers to that above steps requires to carry out the hsv color space according to mode of operation setting or that adjust in real time respectively and carries out subspace division processing, obtains the target image of intending carrying out color classification, the color classification of determining described part or all of pixel, and the color classification of determining described target image.

Preferably, in step S301, sorting device 1:

For example, in step S301, sorting device 1 carries out the subspace to the hsv color space and divides processing by machine learning, and to obtain the red color subspace, the step of machine learning is as follows:

Step 3: in described every sub spaces, set first initial red curve, such as the red curve formula be: f1 (H, S)=H+S^a1*b1+c1, f2 (H, S)=H+S^a2*b2+c2, wherein a, b, c is parameter, a, b, the span of c is respectively: a (1/,100 50), b (360 360), c (360 360), the HSV value of target image pixel drops on the redness that is labeled as between two curves;

More preferably, in step S301, sorting device 1:

For example the user is only interested in scarlet color, and presetting the color classification number is 1, and in step S301, sorting device 1 is carried out above-mentioned steps 1 to 8 according to the color classification number, to obtain the corresponding color sub-spaces of scarlet color.

It is 1 by presetting the color classification number that those skilled in the art can understand above-mentioned; by machine learning; take the method that obtains the described a plurality of color sub-spaces corresponding with described color classification number only as giving an example; other are existing or may occur from now on according to the color classification number that presets; by machine learning; the subspace is carried out in the hsv color space divide processing; with the method that obtains the described a plurality of color sub-spaces corresponding with described color classification number as applicable to the present invention; also should be included in the protection domain of the present invention, and be contained in this with way of reference.

Preferably, in step S304, sorting device 1:

For example, in step S304, sorting device 1 is according to the color classification of each pixel, and the weight of each pixel, and the color classification of described target image is determined in weighting.Perhaps, for example, determine the color classification of described target image by the statistical information weighting of color classification, wherein red weights are 0.1, golden weights are 0.4, green weights are 0.4, and blue weights are 0.1, and the statistical information of the pixel color classification of target image is: redness accounts for 50%, gold accounts for 20%, green accounts for 20%, and blueness accounts for 10%, in step S304, sorting device 1 is according to statistical information and the weights of the pixel color classification of target image, the statistical information of the pixel of the target image after the calculating weighting is: redness accounts for 22.7%, and gold accounts for 36.3%, and green accounts for 36.3%, blueness accounts for 4.5%, in step S304, sorting device 1 is according to the pixels statistics information of the target image after the weighting, and the color classification of determining described target image is gold+green.

More preferably, in step S304, sorting device 1:

For example, by the color classification that described target image is determined in statistical information and the band of position weighting of pixel in described target image of color classification, a width of cloth target image is square, its picture material is an one-by-one inch photograph, and character image mainly concentrates on center section, uses the straight line perpendicular to the base that the target image draw is divided into 3 zones, be respectively left, in, the right side, wherein yellow on a left side, in, right weights are respectively 0.1,0.8,0.1, black is on a left side, in, right weights are respectively 0.1,0.8,0.1, blue weights in left, middle and right are respectively 0.1,0.8,0.1 the statistical information of the pixel color classification of target image is: yellow accounts for 30%, and black accounts for 30%, blueness accounts for 40%, wherein yellow accounts for respectively 10% in left, middle and right, 80%, 10%, black is on a left side, in, part on the right side and do not account for 10%, 80%, 10%, blue on a left side, in, part on the right side and do not account for 40%, 20%, 40%, in step S304, sorting device 1 is according to the statistical information of the pixel color classification of target image and the weights of present position, and the statistical information of the target image pixel after the calculating weighting is: yellow accounts for 40.2%, and black accounts for 40.2%, blueness accounts for 19.6%, in step S301, sorting device 1 is according to the pixels statistics information of the target image after the weighting, and the color classification of determining described target image is yellow+black.

Preferably, sorting device 1 also is included in step S311 (not shown), and in step S311,1 pair of described target image of sorting device carries out pre-service, to obtain pretreated described target image.

Wherein, the pretreatment operation of described step S311 comprise following at least each:

-described target image is compressed processing;

-described target image is intercepted processing.

For example, become the color space mapping in hsv color space to process the RGB color space conversion of target image, pixel s in the objective definition image is (R, G, B) in the assignment of rgb space, wherein distinguish R, G, B represent respectively red, green and blue assignment, the span of RGB has been done normalization, it is the real number between 0 to 1, if max is R, G, the maximum among the B, min is R, G, the reckling among the B is in step S311, each pixel of 1 pair of described target image of sorting device is carried out color space mapping processing, and the RGB color space is as follows to the mapping treatment method in hsv color space:

max＝max(R，G，B)

min＝min(R，G，B)

ifR＝max，H＝(G-B)/(max-min)

ifG＝max，H＝2+(B-R)/(max-min)

ifB＝max，H＝4+(R-G)/(max-min)

H＝H*60

ifH＜0，H＝H+360

V＝max(R，G，B)

S＝(max-min)/max

More preferably, in step S303, sorting device 1:

For example, in step S311, sorting device 1 converts the RGB color space of target image e to the hsv color space through color space mapping processing, the HSV[0 of pixel r after wherein processing, 0.5,0.05], the hsv color space uses the space plane vertical with the V axle that the V axle is divided equally into 10 five equilibriums, be about to the sub spaces that described hsv color spatial division becomes 10 sub spaces, the subspace is HSV[0-360 under one of them, 0-1,0-0.1], in step S303, sorting device 1 determines that described pixel r belongs to subspace HSV[0-360,0-1,0-0.1], sorting device 1 shines upon the HSV value of pixel r and the color space of this subspace, determine that this pixel r drops between red curve f1 and the f2, and then determine that this pixel r belongs to red, in step S303, sorting device 1 can adopt the mode of batch processing or the mode of distributed treatment that part or all of pixel in the target image is shone upon, to determine the color classification of described part or all of pixel.

Fig. 4 illustrate in accordance with a preferred embodiment of the present invention by computer implemented method flow diagram for target image being carried out color classification, wherein, the method also comprises step S405.Referring to Fig. 4 the preferred embodiment is described: particularly, in step S401,1 pair of hsv color space of sorting device is carried out the subspace and is divided processing, to obtain a plurality of color sub-spaces, wherein, the corresponding a kind of color of each color sub-spaces; In step S402, sorting device 1 obtains with the user by the corresponding described target image of the search sequence of subscriber equipment input; In step S403, sorting device 1 is according to the HSV assignment of described target image under described hsv color space, determine in the described target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel; In step S404, sorting device 1 is determined the color classification of described target image according to the color classification of described part or all of pixel; In step S405, sorting device 1 offers described subscriber equipment with the color classification of described target image and described target image.Wherein, step S401, step S402, step S403 and step S404 are same or similar with corresponding step shown in Figure 3 respectively, so locate to repeat no more, and mode by reference is contained in this.Wherein, described subscriber equipment comprises PDA, mobile phone terminal, computing machine, has the digital terminal of network savvy etc.

Particularly, in step S402, sorting device 1 obtains with the user by the corresponding described target image of the search sequence of subscriber equipment input.For example, the user uses mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", in step S402, sorting device 1 uses the HTTPs agreement, calls the API that the network picture database provides, and obtains the target image relevant with " blue Garfield ".

In step S405, sorting device 1 offers described subscriber equipment with the color classification of described target image and described target image.For example, the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", in step S402, sorting device 1 uses the HTTPs agreement, calls the inquiry API that the network picture database provides, obtain 1000 target images relevant with " blue Garfield ", its storage format is the picture of jpg; In step S403, sorting device 1 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; In step S404, sorting device 1 is determined the color classification of described target image according to the color classification of described whole pixels; In step S405, sorting device 1 offers described subscriber equipment with the color classification of described target image and described target image, such as the form of color classification by form, and the mode of figure, target image is by the form of file, the form of webpage etc.

Preferably, in step S405, sorting device 1 offers described subscriber equipment according to the color classification of described target image with described target image.

For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", in step S402, sorting device 1 uses the HTTPs agreement, call the API that the network picture database provides, obtain the image of 1000 " Garfields ", its storage format is the picture of jpg; In step S403, sorting device 1 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; In step S404, sorting device 1 is determined the color classification of described target image according to the color classification of described whole pixels; In step S405, sorting device 1 is according to the color classification of described target image, described target image is offered described subscriber equipment, generate different files such as target image according to color classification, target image presents at the diverse location of webpage according to color classification etc.

Preferably, the method also comprises step S406 and step S407, and wherein, in step S406, sorting device 1 obtains the corresponding described user of described search sequence to the color demand information of described target image by semantic analysis; In step S407, sorting device 1 is according to color classification and the described color demand information of described target image, the matching image of selecting described color classification and described color demand information to be complementary from described target image; In step S405, sorting device 1 offers described subscriber equipment with described matching image.

Particularly, in step S406, sorting device 1 obtains the corresponding described user of described search sequence to the color demand information of described target image by semantic analysis.Wherein, semantic analysis comprises that the method for using statistical computation analyzes a large amount of text sets, thereby extract semantic structure potential between word and the word, and with this potential semantic structure, obtain the corresponding described user of described search sequence to the color demand information of described target image.

For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield " is in step S406, it is Garfield that sorting device 1 obtains user's query demand by semantic analysis, and color is blue.

In step S407, sorting device 1 is according to color classification and the described color demand information of described target image, the matching image of selecting described color classification and described color demand information to be complementary from described target image.

For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", in step S406, it is Garfield that sorting device 1 obtains user's query demand by semantic analysis, color is blue, in step S402, sorting device 1 uses the HTTPs agreement, calls the API that the network picture database provides, obtain the image of 1000 " Garfields ", its storage format is the picture of jpg; In step S403, sorting device 1 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; In step S404, sorting device 1 is determined the color classification of described target image according to the color classification of described whole pixels; In step S407, sorting device 1 is selected described color classification and the blue Garfield image that is complementary according to color classification and the described color demand information of described target image from described target image.

In step S405, sorting device 1 offers described subscriber equipment with described matching image.

For example the user passes through mobile phone terminal, be connected to the internet by the WiFi mode, input text Query Information in the cell-phone customer terminal of network picture database " blue Garfield ", in step S406, it is Garfield that sorting device 1 obtains user's query demand by semantic analysis, color is blue, in step S402, sorting device 1 uses the HTTPs agreement, calls the API that the network picture database provides, obtain the image of 1000 " Garfields ", its storage format is the picture of jpg; In step S403, sorting device 1 is according to the HSV assignment of target image under described hsv color space of obtaining, and determines in the described target image all mapping relations of pixels and described a plurality of color sub-spaces, with the color classification of definite described whole pixels; In step S404, sorting device 1 is determined the color classification of described target image according to the color classification of described whole pixels; In step S407, sorting device 1 is according to color classification and the described color demand information of described target image, selecting described color classification from described target image is the Garfield image that is complementary with blueness, in step S405, sorting device 1 offers described subscriber equipment with described matching image.

To those skilled in the art, obviously the invention is not restricted to the details of above-mentioned example embodiment, and in the situation that do not deviate from spirit of the present invention or essential characteristic, can realize the present invention with other concrete form.Therefore, no matter from which point, all should regard embodiment as exemplary, and be nonrestrictive, scope of the present invention is limited by claims rather than above-mentioned explanation, therefore is intended to be included in the present invention dropping on the implication that is equal to important document of claim and all changes in the scope.Any Reference numeral in the claim should be considered as limit related claim.In addition, obviously other unit or step do not got rid of in " comprising " word, and odd number is not got rid of plural number.A plurality of unit of stating in the device claim or device also can be realized by software or hardware by a unit or device.The first, the second word such as grade is used for representing title, and does not represent any specific order.

Claims

1. one kind by computer implemented method for target image being carried out color classification, and wherein, the method may further comprise the steps:

Wherein, the method also comprises:

A obtains the target image of intending carrying out color classification;

2. method according to claim 1, wherein, described step x also comprises:

3. method according to claim 2, wherein, described step x also comprises:

4. each described method in 3 according to claim 1, wherein, described step c also comprises:

5. method according to claim 4, wherein, described step c also comprises:

6. each described method in 5 according to claim 1, wherein, the method also comprises:

H carries out pre-service to described target image, to obtain pretreated described target image;

Wherein, described step b also comprises:

7. method according to claim 6, wherein, the pretreatment operation among the described step h comprise following at least each:

-described target image is compressed processing;

-described target image is intercepted processing.

8. each described method in 7 according to claim 1, wherein, described step a also comprises:

-obtain with the user by the corresponding described target image of the search sequence of subscriber equipment input;

Wherein, the method also comprises:

R offers described subscriber equipment with the color classification of described target image and described target image.

9. method according to claim 8, wherein, described step r also comprises:

-according to the color classification of described target image, described target image is offered described subscriber equipment.

10. according to claim 8 or 9 described methods, wherein, the method also comprises:

-by semantic analysis, obtain the corresponding described user of described search sequence to the color demand information of described target image;

-according to color classification and the described color demand information of described target image, the matching image of from described target image, selecting described color classification and described color demand information to be complementary;

Wherein, described step r also comprises:

-described matching image is offered described subscriber equipment.

11. a sorting device that is used for target image is carried out color classification, wherein, this equipment comprises:

Wherein, this equipment also comprises:

12. equipment according to claim 11, wherein, described spatial division device also is used for:

13. equipment according to claim 12, wherein, described spatial division device also is used for:

14. each described equipment in 13 according to claim 11, wherein, described color determines that device also is used for:

15. equipment according to claim 14, wherein, described color determines that device also is used for:

16. each described equipment in 15 according to claim 11, wherein, this equipment also comprises:

Pretreatment unit is used for described target image is carried out pre-service, to obtain pretreated described target image;

Wherein, described spatial mappings device also is used for:

-according to the HSV assignment of described pretreated target image under described hsv color space, determine in the described pretreated target image the partly or entirely mapping relations of pixel and described a plurality of color sub-spaces, with the color classification of definite described part or all of pixel;

17. equipment according to claim 16, wherein, the pretreatment operation in the described pretreatment unit comprise following at least each:

-described target image is compressed processing;

-described target image is intercepted processing.

18. each described equipment in 17 according to claim 11, wherein, described image acquiring device also is used for:

Wherein, this equipment also comprises:

Image providing device is used for the color classification of described target image and described target image is offered described subscriber equipment.

19. equipment according to claim 18, wherein, described image providing device also is used for:

20. according to claim 18 or 19 described equipment, wherein, this equipment also comprises:

The Requirement Acquisition device is used for by semantic analysis, obtains the corresponding described user of described search sequence to the color demand information of described target image;

Image matching apparatus is used for color classification and described color demand information according to described target image, the matching image of selecting described color classification and described color demand information to be complementary from described target image;

Wherein, described image providing device also is used for described matching image is offered described subscriber equipment.

21. a browser comprises: such as each described image classification device that target image is carried out color classification in the claim 11 to 20.

22. a search engine comprises: such as each described image classification device that target image is carried out color classification in the claim 11 to 20.