CN1685365A - image retrieval program - Google Patents
image retrieval program Download PDFInfo
- Publication number
- CN1685365A CN1685365A CNA02829713XA CN02829713A CN1685365A CN 1685365 A CN1685365 A CN 1685365A CN A02829713X A CNA02829713X A CN A02829713XA CN 02829713 A CN02829713 A CN 02829713A CN 1685365 A CN1685365 A CN 1685365A
- Authority
- CN
- China
- Prior art keywords
- image
- view data
- similarity
- characteristic quantity
- data
- 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
- 238000004364 calculation method Methods 0.000 claims abstract description 15
- 239000000284 extract Substances 0.000 claims abstract description 7
- 238000000605 extraction Methods 0.000 claims description 13
- 238000013075 data extraction Methods 0.000 abstract description 4
- 238000000034 method Methods 0.000 description 35
- 238000012876 topography Methods 0.000 description 25
- 230000008569 process Effects 0.000 description 11
- 239000006185 dispersion Substances 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000010354 integration Effects 0.000 description 3
- 230000035807 sensation Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 206010002091 Anaesthesia Diseases 0.000 description 1
- 241001062009 Indigofera Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000001949 anaesthesia Methods 0.000 description 1
- 230000037005 anaesthesia Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000016776 visual perception Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
- G06F16/532—Query formulation, e.g. graphical querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Library & Information Science (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Processing Or Creating Images (AREA)
Abstract
一种图像检索程序。具有：特征量数据抽出部(103)，从关键字图像数据中抽出多个类型(例如，颜色配置特征量、纹理特征量、形状特征量)的特征量数据；适合性判断部(104)，对于特征量数据抽出部(103)所抽出的各特征量数据的类型，把是否计算出与人的类似感觉相符合的相似度作为有无适合性来进行判断；相似度计算部(105)，根据适合性判断部(104)的判断结果，计算关键字图像数据与检索对象的各图像数据的相似度；检索部(106)，把与相似度对应的图像数据作为检索结果输出。
An image retrieval program. Have: feature quantity data extraction part (103), extract the feature quantity data of a plurality of types (for example, color configuration feature quantity, texture characteristic quantity, shape characteristic quantity) from keyword image data; Suitability judging part (104), For the type of each feature quantity data extracted by feature quantity data extracting part (103), judge whether to calculate the similarity matching with people's similar feeling as having suitability; Similarity calculation part (105), According to the judgment result of the suitability judging part (104), the similarity between the keyword image data and each image data of the retrieval object is calculated; the retrieval part (106) outputs the image data corresponding to the similarity as the retrieval result.
Description
Technical field
The present invention relates to according to the key word image, from image data base, retrieve the image retrieval program of desirable view data, particularly can improve the image retrieval program of retrieval precision.
Background technology
Up till now, developed the image retrieving apparatus that much has following function, promptly, by from the image data base that has stored the great amount of images data, taking out in order and also showing, make the user can find desirable view data efficiently as the view data of retrieving the key word image similarity that the user gave that requires.
In such image retrieving apparatus, used the retrieving similar images technology, owing to calculate the similarity of the view data of key word view data and searching object with numerical value, so defined the characteristic quantity that comes the presentation video data characteristics with numerical value.In calculating, the similarity of image uses this characteristic quantity.
Here, in above-mentioned characteristic quantity, the type that the viewpoint of the features such as shape of the profile of existing object is divided in the configuration of the color that with good grounds seizure image had, texture (be meant and be used for being illustrated in the pattern on surface and the describing of texture that figure etc. is attached to figure), the image.
Like this, in the past, to the type of each characteristic quantity is that to feel consistent be target with relevant with people's similarity, (record and give birth in the sheet mountain, Sato Shinichi: the index technology that is used to carry out similar to search to have proposed a lot of effectively definition and its extraction method, information processing association magazine, Vol.42, No.10, pp.958-964 (2001) .).
Therefore, in image retrieving apparatus in the past, owing to retrieve similar image according to various viewpoints, thereby will prepare the characteristic quantity of a plurality of types in advance, after having specified the type of the characteristic quantity that meets searched targets, implement retrieval again by the user.
But in image retrieving apparatus in the past, for the general user who does not possess the knowledge relevant with the characteristic quantity of image, the type that select to meet the characteristic quantity of searched targets is very very difficult.
Consequently, need carry out retrieval repeatedly, after through retrieval failure repeatedly, just can determine the correct characteristic quantity type that is used to retrieve, thus, the problem that search operaqtion needs the very big workload of user can occur according to the characteristic quantity type of tentative decision.
And, relevantly with similarity feel consistent for what use that similarity that these characteristic quantities calculated had with the people, the content that becomes the view data of object just must satisfied certain condition to each characteristic quantity type defined.
Therefore, include in searching object under the situation of ineligible view data, the feeling of similarity calculating meeting and people disagreed, and includes for the user the obviously view data of the similarity of anaesthesia and key word image in the result for retrieval.Under only can being restricted to, can not produce this problem to situation about retrieving with the corresponding to view data of condition of the type of the characteristic quantity of the view data of using searching object.
But, under the situation of image data set as searching object of the view data of various content that will be mixed with the collected view data in internet that use popularizes in recent years etc., the problem that retrieval precision descends will appear in the inconsistent view data of condition that comprises the type of a lot of and characteristic quantity.
The present invention proposes in view of the above-mentioned problems, and its purpose is to provide the image retrieval program that can improve retrieval precision.
Summary of the invention
To achieve the above object, the present invention is a kind of image retrieval program, and it makes the function of the following functional unit of computing machine performance, that is: characteristic quantity is extracted the unit out, extracts the characteristic quantity of a plurality of types from the key word image out; The adaptability judging unit, the type of extracting each characteristic quantity of extracting out the unit out for described characteristic quantity feels that whether calculating to the similar of people corresponding to similarity is as having or not adaptability to judge; Similarity calculated according to the judged result of described adaptability judging unit, is calculated the similarity of each image of described key word image and searching object; The result for retrieval output unit is exported the image corresponding with described similarity as result for retrieval.According to the present invention, type for each characteristic quantity of from the key word image, being extracted out, feel that with whether calculating corresponding to similarity is as having or not adaptability to judge to the similar of people, calculate the similarity of each image of key word image and searching object according to this judged result, to export as result for retrieval with the corresponding image of similarity, thereby can prevent from the selection mistake of the type of employed characteristic quantity when calculating similarity to improve retrieval accuracy.
Description of drawings
Fig. 1 is the block scheme of the image retrieving apparatus of expression first embodiment of the present invention.
Fig. 2 is the figure of the list structure of expression image data base 110 shown in Figure 1.
Fig. 3 is the figure of the list structure of expression characteristic quantity database 120 shown in Figure 1.
Fig. 4 is the process flow diagram of the action of the identical embodiment of explanation.
Fig. 5 is the process flow diagram that explanation characteristic quantity data storing shown in Figure 4 is handled.
Fig. 6 is the process flow diagram of explanation retrieval process shown in Figure 4.
Fig. 7 is the process flow diagram of explanation adaptability judgment processing shown in Figure 6.
Fig. 8 is the figure of explanation retrieval process shown in Figure 6.
Fig. 9 is the figure that the adaptability relevant with the color configuration characteristic quantity of the same embodiment of explanation judged.
Figure 10 is the figure that the adaptability relevant with the color configuration characteristic quantity of the same embodiment of explanation judged.
Figure 11 is the figure that the adaptability relevant with the color configuration characteristic quantity of the same embodiment of explanation judged.
Figure 12 is the figure of the color configuration characteristic quantity of the same embodiment of explanation.
Figure 13 is the figure of employed computing formula during the adaptability relevant with the color configuration characteristic quantity of the same embodiment of expression judged.
Figure 14 is the figure that the adaptability relevant with texture characteristic amount of the same embodiment of explanation judged.
Figure 15 is the figure that the adaptability relevant with the shape facility amount of the same embodiment of explanation judged.
Figure 16 is the figure that is illustrated in the result for retrieval picture 300 under the situation of having used the color configuration characteristic quantity in the same embodiment.
Figure 17 is the figure that is illustrated in the result for retrieval picture 310 under the situation of having used texture characteristic amount in the same embodiment.
Figure 18 is the figure that is illustrated in the result for retrieval picture 320 under the situation of having used the shape facility amount in the same embodiment.
Figure 19 is the block scheme of structure of the variation of the same embodiment of expression.
Embodiment
Below, be described with reference to the accompanying drawings the image search method and the image retrieving apparatus of first embodiment of the present invention.Fig. 1 is the block scheme of the image retrieving apparatus of expression first embodiment of the present invention.
This image retrieving apparatus 100 is in the middle of a plurality of characteristic quantities (color configuration characteristic quantity, texture characteristic amount, shape facility amount etc.) that use in order to calculate the similarity between the view data in the described image data base 100 in key word view data and back, select the similar characteristic quantity of feeling that corresponding to possibility is high to the people, and use this characteristic quantity to carry out image retrieval.
In image retrieving apparatus 100, input part 101 is in the input of the key word view data of the search key that becomes retrieving similar images, reaches employed input media in the various key word inputs, is digital camera, scanner, internet communication machine, keyboard, mouse etc.
For example, image I D is the view data of " 0000001 ", with the view data 111 of the searching object shown in Fig. 8 (b)
1Corresponding.This view data 111
1The wallpaper image of expression longitudinal grin pattern.
In addition, image I D is the view data of " 0000002 ", with the view data 111 of the searching object shown in Fig. 8 (c)
2Corresponding.This view data 111
2With above-mentioned view data 111
1The same, the wallpaper image of expression longitudinal grin pattern.
In addition, image I D is the view data of " 0000003 ", with the view data 111 of the searching object shown in Fig. 8 (d)
3Corresponding.This view data 111
3The trees of forest and the photograph image in house have been taken in expression.
Here, in the first embodiment, as the type of the characteristic quantity of image, illustration three kinds of following (1)～(3).
(1) color configuration characteristic quantity
(2) texture characteristic amount
(3) shape facility amount
Above-mentioned (1) color configuration characteristic quantity is the characteristic quantity of the space distribution state of the color on the presentation video data.(2) texture characteristic amount is the pattern of presentation video data and the characteristic quantity of describing of texture.(3) the shape facility amount is the characteristic quantity of the contour shape of the object that exists in the presentation video data.
Specifically, as shown in Figure 3, characteristic quantity database 120 is made of color configuration characteristic quantity tables of data 121, texture characteristic amount tables of data 122 and shape facility amount tables of data 123.
These color configuration characteristic quantity tables of data 121, texture characteristic amount tables of data 122 and shape facility amount tables of data 123 have the field of image I D and characteristic quantity data respectively.Image I D is corresponding with the image I D of image data base 110 (with reference to Fig. 2).
The characteristic quantity data of color configuration characteristic quantity tables of data 121 are expression and data corresponding to the relevant color configuration characteristic quantity of the view data of image I D.The characteristic quantity data of texture characteristic amount tables of data 122 are expression and data corresponding to the relevant texture characteristic amount of the view data of image I D.The characteristic quantity data of shape facility amount tables of data 123 are expression and data corresponding to the relevant shape facility amount of the view data of image I D.
Turn back to Fig. 1, characteristic quantity data extraction unit 103 is extracted out from each view data of the stored searching object of image data base 110 and key word view data respectively and the corresponding characteristic quantity data of above-mentioned three kinds of characteristic quantities (color configuration characteristic quantity, texture characteristic amount, shape facility amount), and they are stored in the characteristic quantity database 120 (with reference to Fig. 4).
Under the situation of the similarity of the view data of calculating key word view data and searching object, use these characteristic quantity data.
Adaptability judging part 104 is for calculating the key word view data with numerical value and employed characteristic quantity data during as the similarity between the view data of searching object, under the situation of having carried out calculation of similarity degree, have or not adaptability to judge as the characteristic quantity data (characteristic quantity) and the key word view data (view data) of being given with whether having calculated the corresponding to similarity of similar sensation (similarity) that had to the people.
Similarity calculating part 105 according to and be judged as adaptability the key word view data the corresponding vector value of characteristic quantity data and and the corresponding vector value of characteristic quantity data of each view data of searching object between Euclidean distance, calculate the similarity of each view data of searching object to the key word view data.Search part 106 is carried out the processing relevant with image retrieval.
Below, the action to first embodiment describes with reference to the process flow diagram of Fig. 4～shown in Figure 7.In image data base shown in Figure 1 110, as shown in Figure 2, store a plurality of view data of searching object in advance.
Under this state, in step SA1 shown in Figure 4, judge to have or not that in this case, judged result is a "No" from the relevant requirement of user's storage with the characteristic quantity data.
In step SA2, judge the requirement relevant that has or not from the user with image retrieval, in this case, judged result is a "No".Then, the judgement of repeating step SA1 and step SA2 is a "Yes" up to the judged result of step SA1 or step SA2.
Then, as from user's storage with the characteristic quantity data relevant require the time, the judged result of step SA1 is a "Yes".In step SA3, from the stored view data of image data base 110, extract out and the corresponding characteristic quantity data of foregoing three kinds of characteristic quantities (color configuration characteristic quantity, texture characteristic amount, shape facility amount), and implement the characteristic quantity data storing that they are stored in the characteristic quantity database 120 are handled.
Particularly, in step SB1 shown in Figure 5, characteristic quantity extraction unit 103 obtains a view data from image data base 110 (with reference to Fig. 2) (for example, view data 111
1(with reference to Fig. 8 (b)).In step SB2, characteristic quantity extraction unit 103 is extracted out from the view data that obtains among step SB1 respectively and the corresponding characteristic quantity data of three kinds of characteristic quantities (color configuration characteristic quantity, texture characteristic amount, shape facility amount).
Below, to being elaborated with the extraction method of color configuration characteristic quantity, texture characteristic amount, the corresponding characteristic quantity data of shape facility amount respectively.
The color configuration characteristic quantity
At first, extraction method with the corresponding characteristic quantity data of color configuration characteristic quantity is described.
As shown in Figure 9, for example will be with suitable several each partial image data I when view data is divided into clathrate in length and breadth
IjThe mean value of the color value value of arranging as one dimension come apparent color configuration feature amount.
Here, be that the trivector of composition is represented partial image data I in order to the intensity of R (red), the G (green) of RGB color space, B (indigo plant)
IjThe mean value of color value.
In addition, make partial image data I
IjThe mean value of color value be (R
Ij, G
Ij, B
Ij) situation under, with vector (R
11, G
11, B
11, R
12, G
12, B
12, R
44, G
44, B
44) represent color configuration.
In order to make the color configuration characteristic quantity have meaning, two conditions (1A) below satisfying and partial image data (2A) must reach more than the some.
Condition (1A)
The color that ratio comprised in view data more than the with good grounds fixed value.
Condition (2A)
Having the color pixel that has satisfied condition (1A) spatially concentrates.
This two conditions (1A) and (2A) be the mean value of color value of expression partial image data and the akin condition of color of this partial image data integral body.
For example, a lot of colors all have only under the situation of a little existence in view data, even obtain the mean value of its color value, can not become good approximate value.
In addition, even in the color of the ratio of image data memory more than fixed value, this color is being divided under the situation that thin zone disperses, the also imperceptible significant color of people.Therefore, as condition, pixel will spatially be concentrated in view data.
Judging whether to have satisfied above-mentioned condition (1A) according to following method reaches (2A).
The determination methods of condition (1A)
At first, the RGB color space is divided into the part color space.Then, after the number of the pixel that partial image data comprised of having calculated the each several part color space, calculate the ratio value of total pixel number of the corresponding topography of pixel count of result of calculation by each part color space.
Have at this ratio value under the situation of the part color space more than the setting, with the color value of this part color space representative as representative color to topography that should topography.
As the determining method of the representative color of topography, for example handlebar is positioned at the color of center of pairing part color space as the method for representative color.Under the situation of the representative color that has this topography, just satisfied condition (1A).
The determination methods of condition (2A)
For the representative color of the topography that in the deterministic process of condition (1A), obtains, calculate concentration degree with following method.At first, as shown in Figure 12,, use the window that this pixel is concentrated to the center of the size of w * h, several RC of the corresponding pixel of representative color of the topography that exists among calculating and this window m for all pixels on the M of topography
Xy
At this moment, calculate concentration degree SC according to formula shown in Figure 13 (1).In formula (1), RC is and the sum of the corresponding pixel of representative color of topography that N is the value of regulation.
Have at this concentration degree SC under the situation of representative color of the topography more than the setting, just satisfied condition (2A).
Texture characteristic amount
Below, the extraction method with the corresponding characteristic quantity data of texture characteristic amount is described.
When extracting out, accounted for major part by the situation of identical texture (decorative pattern) prerequisite that is covered as with view data integral body with the corresponding characteristic quantity data of texture characteristic amount.Therefore, under calculating not by the situation of the identical view data that texture covered, these computing method just may be taken out the characteristic quantity data of thinking as expected.
Its result, when image retrieval, when having provided not by identical key word view data that texture covered, among the view data of searching object, can occur and compare, do not feel that according to people's sensation similar view data has showed the situation of higher similarity on the contrary according to the similar view data of feeling as people's sensation.
Come in proper order under the situation of display image result for retrieval at height by similarity and since according to the people feel feel and dissimilar view data on the contrary than feeling that the image similar to the key word view data is more forward, thereby just make recall precision descend.
For example, as the key word image of image retrieval, provided the key word view data 200 shown in Fig. 8 (a).On the other hand, searching object is the view data 111 shown in Fig. 8 (b), (c) reach (d)
1, 111
2And 111
3
Key word view data 200 and view data 111
3It is the photograph image of having taken the landscape that the house is arranged in trees.On the other hand, view data 111
1And 111
2Corresponding to by the identical wallpaper image that texture covered.
In addition, method (H.Tamura according to well-known Tian Cun, S.Mori, andT.Yamawaki, " Texture Features Corresponding Visual Perception, " IEEETrans.System Man and Cybernetics, Vol.8, No.6, (1978) .) calculate texture characteristic amount.
Method according to Tian Cun, assign to represent texture characteristic amount with " roughness (coarseness) ", " contrast (contrast) ", " directivity (directionality) " these three one-tenth, the degree of each composition is extracted out from view data as the trivector with numeric representation respectively.
" roughness " is the size of the scale of the pattern that showed in the presentation video data, and scale is big more, and the value of " roughness " is just big more." contrast " is the unbalanced degree of expression brightness value, and unbalanced degree is big more, and " contrast " value is just big more.
The degree that the direction of the marginal element in " directivity " presentation video data is concentrated to fixed-direction, the frequency of the direction that the frequency in the direction of marginal element is the highest is big more, and " directivity " value is just big more.
Method according to this Tian Cun, from key word view data 200[with reference to Fig. 8 (a)] extracted out with the corresponding characteristic quantity data conditions of texture characteristic amount under, in any one of " roughness " of key word view data 200, " contrast ", " directivity ", compared with view data 111
3, view data 111
1, view data 111
2Represented higher similarity.
But, in above-mentioned image retrieval, if the user search landscape image comprises view data 111 in result for retrieval
1And view data 111
2Just imappropriate.Because the ill-considered view data of this kind is positioned at the forward position of result for retrieval, thereby, will see a lot of unnecessary view data, cause the low of recall precision in order to find destination image data.
Therefore, for such key word view data, calculate the homogeny of texture, under the situation of value less than the value of regulation of homogeny, being judged as does not have adaptability, can not use texture characteristic amount in retrieval.
Computing method to the homogeny of texture describe.At first, the image of calculating object is cut apart in length and breadth.For example, shown in Figure 14 (a) and Figure 14 (B), view data (key word view data 200 etc.) four is cut apart by halving in length and breadth.
Then, from each topography of being cut apart, textural characteristics is extracted out as vector.As the computing method of textural characteristics of this moment, can be the identical method of method with employed calculated characteristics during similarity is calculated, also can be method for distinguishing.
Here, be the method for Tian Cun if use identical method, from each topography, extract the feature of trivector out.The homogeny that can represent texture here, according to the unbalanced degree of the eigenvector of being extracted out.
Identical, unbalanced will be more little.Therefore, to from resulting four eigenvectors of topography, calculate dispersion value as unbalanced degree, under the situation of dispersion value greater than the value of regulation, homogeny is just low, and promptly being judged as does not have adaptability.In the example of Figure 14 (a), under the situation of key word view data 200, because the topography of bottom right and other three compare, in " roughness ", " contrast ", " directivity " any one all has a great difference, thereby dispersion value just uprises, and can be judged as does not have adaptability.
The shape facility amount
Below, the extraction method with the corresponding characteristic quantity data of shape facility amount is described.
With the extraction of the corresponding characteristic quantity data of shape facility amount in, method under the situation of extracting outline line from view data in advance out is arranged, do not extract the method under the situation of outline line out, but under the situation of handling collected view data from the internet, be not in the great majority owing to do not extract the situation of outline line in advance out, thereby use the method for back.
On the other hand, be under the situation of object with view data arbitrarily, because it is very difficult technically to extract the outline line of the object that exists in the view data out, thereby just utilize the outline line of object in view data, to become very strong marginal element, distribute as the different frequency of direction of the marginal element of part and represent the shape facility amount approx.
In the extraction method of planting the corresponding characteristic quantity data of shape facility amount therewith, be prerequisite with following conditions (3A).
Condition (3A): be the background of monochrome, have object at the specific part of view data.
As specific part, for example near the center of view data.
Here, the method for judging based on the adaptability of condition (3A) is as described below.At first, shown in Figure 15 (a) and Figure 15 (b), along from the center of view data by the angle of each regulation with radial line segment of being pulled out, on the direction at the center of view data, pixel is being scanned.
In the process of scanning, calculate the difference of the brightness value of continuous pixel successively, store the coordinate figure (x, y) of the point that converges mutually with the difference of the value that surpasses regulation.
For from each scan lines being carried out the resulting coordinate figure of result of this processing, the distance of the coordinate figure between the adjacent scan lines is added up.With this aggregate-value is benchmark, and under the situation greater than the value of stipulating, being judged as does not have adaptability.Under the situation of the view data that satisfies precondition 210 as Figure 15 (a), the distance between the stored coordinate figure just diminishes.
On the other hand, for the key word view data 200 shown in Figure 15 (b), in object (house) background in addition, also describe other object, from such image, extracting out under the situation of shape facility, because the influence of the outline line beyond the object etc. is very big, thereby can not obtain significant shape facility.
Under the situation of this view data,, thereby can be judged as according to this determination methods and do not have adaptability because that the distance between the stored coordinate figure relatively becomes is big.
Turn back to Fig. 5, in step SB3, each characteristic quantity data that characteristic quantity extraction unit 103 will be extracted out in step SB2 (color configuration characteristic quantity, texture characteristic amount and shape facility amount) are stored into (with reference to Fig. 3) in the characteristic quantity database 120.
In step SB4, characteristic quantity extraction unit 103 judges whether all view data of image data base 110 have been carried out handling (extraction of characteristic quantity, storage), and in this case, judged result is a "No".Then, remaining view data is carried out above-mentioned processing.
Then, when processing finished, the judged result of step SB4 was a "Yes", and the processing of characteristic quantity data storing finishes.
Then, as from the requiring of user's relevant image retrieval the time, the judged result of step SA2 shown in Figure 4 is a "Yes" just.In step SA4, carry out the retrieval process that is used to retrieve the view data similar to the key word view data.
Particularly, in step SC1 shown in Figure 6, the user imports for example key word view data 200 shown in Fig. 8 (a) by input part 101.This key word view data 200 is corresponding to from the view data 111 shown in different angle shots and Fig. 8 (a)
3The photograph image of the landscape in identical place.
In step SC2, key word view data 200 carried out be used for adaptability judgment processing that the adaptability of each characteristic quantity of color configuration characteristic quantity, texture characteristic amount and shape facility amount is judged.
Particularly, in step SD1 shown in Figure 7, adaptability judging part 104 is judged the adaptability of color configuration characteristic quantity.Promptly, 104 pairs of key word view data 200 of adaptability judging part are cut apart in length and breadth.
, as shown in Figure 9, key word view data 200 is indulged 4 five equilibriums, horizontal 4 five equilibriums here, be divided into 16 partial image data (I
11～I
44).
Next, adaptability judging part 104 is to the partial image data (I of each key word view data 200
11～I
44) calculate the ratio value of the pixel count that each several part color space that R, G, B with the RGB color space carry out timesharing such as 4 respectively comprised, the ratio value of the part color space with maximum scale value as value shown in Figure 10, is asked this value.
Here, ratio value is got the scope of (0.0,1.0), and it is just big more to be worth more vast scale.
In addition, for the color configuration characteristic quantity, when the threshold value of the ratio value of topography's representative color of the determination methods that is used for foregoing condition (1A) was decided to be 0.3, all topographies all had the part color space of the above ratio value of threshold value, have the representative color of topography.
In addition, the N of formula (1) (with reference to Figure 13) of concentration degree SC of determination methods that will be used to obtain the condition (2A) of color configuration characteristic quantity is decided to be 6, and the result of calculation of the concentration degree SC of topography's representative color of ratio value with Figure 10 just is value shown in Figure 11.Concentration degree SC gets the scope of (0.0,1.0), is worth big more just concentrated more.
Here, when the threshold value with the concentration degree SC of the determination methods of the condition (2A) of color configuration characteristic quantity was decided to be 0.6, the topography of other beyond the topography of bottom right shown in Figure 11 had the above concentration degree of threshold value, so satisfy condition (2A).The topography of this bottom right can not satisfy condition (2A).
At last, at the benchmark that will satisfy the adaptability of judging the color configuration characteristic quantity simultaneously is that the threshold value of topography's number of condition (1A) and condition (2A) is decided to be under 14 the situation, because other 15 topographies of key word view data 200 except the topography of bottom right shown in Figure 11 satisfy condition (1A) and condition (2A) simultaneously, thereby are judged as the adaptability that has with the color configuration characteristic quantity.
In step SD2 shown in Figure 7, adaptability judging part 104 is judged the adaptability of texture characteristic amount.That is, shown in Figure 14 (a), adaptability judging part 104 carries out 4 by 2 five equilibriums in length and breadth to key word view data 200 to be cut apart, and from each topography texture characteristic amount is extracted out as vector.
Next, the dispersion value between 104 pairs of each vector values of adaptability judging part for example calculates as 0.7.
Here, dispersion value is got the scope of (0.0,1.0), is worth big more just overstepping the bounds of propriety diffusing.
Here, when the threshold value of the dispersion value of the adaptability that will judge key word view data 200 and texture characteristic amount was decided to be 0.6, the dispersion value of key word view data 200 was more than threshold value.Therefore, in this case,, thereby be judged as the adaptability that does not have with texture characteristic amount because key word view data 200 does not have the homogeny of texture.
In step SD3 shown in Figure 7, adaptability judging part 104 is judged the adaptability of shape facility amount.That is, shown in Figure 15 (b), adaptability judging part 104 along from the center of key word view data 200 by each angle 22.5 degree with radial line segment of being pulled out, on the direction at the center of view data, pixel is being scanned.
Here, adaptability judging part 104 calculates the difference of the brightness value of continuous pixel successively in the process of scanning, store the coordinate figure (x, y) of the pixel of the difference that surpasses the threshold value of stipulating.The threshold value of difference, for example the scope in the value of difference is under the situation of (0.0,1.0), is 0.8.
For from each scan lines being carried out the resulting coordinate figure of result of above processing, the result who the distance of the coordinate figure between the adjacent scan lines is carried out accumulative total is 1150.Here, in the time will being decided to be 1000 to the threshold value of this aggregate-value, key word view data 200 has the aggregate-value bigger than threshold value, and being judged as does not have adaptability to the shape facility amount.
According to the above, for key word view data 200, being judged as with the color configuration characteristic quantity has adaptability, but and texture characteristic amount reaches and the shape facility amount does not have adaptability.
Get back to Fig. 6, in step SC3, characteristic quantity data extraction unit 103 is extracted out from key word view data 200 and the corresponding characteristic quantity data of type (in this case, being above-mentioned color configuration characteristic quantity) that adaptability is arranged.
In step SC4, the characteristic quantity data that characteristic quantity data extraction unit 103 will be extracted out in step SC3 (color configuration characteristic quantity) with and key word view data 200 corresponding image I D (=0000004) be mapped, be stored in the color configuration characteristic quantity tables of data 121 of characteristic quantity database 120 (with reference to Fig. 3).
In addition, in Fig. 3, in texture characteristic amount tables of data 122 and shape facility amount tables of data 123, store corresponding characteristic quantity data, but in fact do not store these data with image I D (=0000004).
In step SC5 shown in Figure 6, similarity calculating part 105 is obtained from color configuration characteristic quantity tables of data 121 shown in Figure 3 and the corresponding characteristic quantity data of image I D (=0000004) (the color configuration characteristic quantity of key word view data 200).These characteristic quantity data are with to be judged as the type (color configuration characteristic quantity) that has with the adaptability of key word view data 200 in step SC2 corresponding.
Then, have or not adaptability between each view data of stored searching object in adaptability judging part 104 judgement the above-mentioned types (color configuration characteristic quantity) and the image data base 110.
Particularly, adaptability judging part 104 is the same with step SD1 shown in Figure 7, judges between above-mentioned color configuration characteristic quantity and each view data to have or not adaptability.
And, be under the situation of texture characteristic amount at the above-mentioned type, adaptability judging part 104 is the same with step SD2 shown in Figure 7, judges between above-mentioned texture characteristic amount and each view data to have or not adaptability.
And, be under the situation of shape facility amount at the above-mentioned type, adaptability judging part 104 is the same with step SD3 shown in Figure 7, judges between above-mentioned shape facility amount and each view data to have or not adaptability.
Then, similarity calculating part 105 is from all view data of searching object, eliminating is judged as the view data that does not have adaptability in above-mentioned adaptability judging part 104, with the searching object set is the view data (for example, with the corresponding view data of image I D (0000001,0000002,0000003)) that is judged as adaptability.
Then, similarity calculating part 105 is obtained the corresponding characteristic quantity data (view data 111 with above-mentioned image I D (=0000001,0000002,0000003)
1, view data 111
2And view data 111
3The color configuration characteristic quantity).
Next, similarity calculating part 105 calculate with key word view data 200 corresponding characteristic quantity data (color configuration characteristic quantity) and respectively with key word view data 111
1, 111
2And 111
3Euclidean distance between the corresponding characteristic quantity data (color configuration characteristic quantity).When second back of radix point rounded up, result's (Euclidean distance) was as follows:
Key word view data 200 and view data 111
1: 111.6
Key word view data 200 and view data 111
2: 101.7
Key word view data 200 and view data 111
3: 7.1
In The above results, Euclidean distance is short more, and the similarity of the view data of key word view data 200 and searching object is just high more.
Therefore, the order to the similarity of key word view data 200 (color configuration characteristic quantity) is: first is view data 111
3(with reference to Fig. 8 (d)), second is view data 111
2(with reference to Fig. 8 (c)), the 3rd is view data 111
1(with reference to Fig. 8 (b)).
In step SC6 shown in Figure 6, search part 106 from image data base 110 (with reference to Fig. 2), obtains view data 111 as result for retrieval according to the order of the similarity of being obtained in step SC5
3, view data 111
2, and view data 111
1
Then, search part 106 shows result for retrieval picture 300 shown in Figure 16 on display part 102.In this result for retrieval picture 300, shown the result for retrieval under the situation of utilizing the color characteristic amount, this figure from left to right shown key word view data 200, by the view data 111 of the height order of similarity
3, view data 111
2And view data 111
1
In addition, in order to understand that corresponding key word view data 200 (picture with scenes) will be felt the most similar view data 111 to the people from result for retrieval picture 300
3(picture with scenes) is presented on the highest position of similarity, will feel dissimilar view data 111
2And view data 111
1Be presented at its back.
Here, suppose that not implementing foregoing adaptability judges, carry out under the situation of image retrieval in utilization and the corresponding characteristic quantity data of texture characteristic amount, use texture characteristic amount tables of data 122 (with reference to Fig. 3) stored characteristic quantity data (texture characteristic amount), key word view data 200 and searching object view data (view data 111
1～111
3) between Euclidean distance be as follows:
Key word view data 200 and view data 111
1: 9.2
Key word view data 200 and view data 111
2: 12.8
Key word view data 200 and view data 111
3: 64.7
Therefore, the order to the similarity of key word view data 200 (texture characteristic amount) is: first is view data 111
1(with reference to Fig. 8 (b)), second is view data 111
2(with reference to Fig. 8 (c)), the 3rd is view data 111
3(with reference to Fig. 8 (d)).
In this case, search part 106 shows result for retrieval picture 310 shown in Figure 17 on display part 102.In this result for retrieval picture 310, shown the result for retrieval under the situation of utilizing texture characteristic amount, this figure from left to right shown key word view data 200, by the view data 111 of the height order of similarity
1, view data 111
2And view data 111
3
In addition, in order to understand that corresponding key word view data 200 (picture with scenes) will be felt dissimilar view data 111 to the people from result for retrieval picture 310
1Be presented on the highest position of similarity, will have felt more similar view data 111
3Be presented on the minimum position of similarity.
In addition, suppose that not implementing foregoing adaptability judges, carry out under the situation of image retrieval in utilization and the corresponding characteristic quantity data of shape facility amount, use stored characteristic quantity data (shape facility amount), key word view data 200 and the searching object view data (view data 111 of shape facility amount tables of data 123 (with reference to Fig. 3)
1～111
3) between Euclidean distance be as follows:
Key word view data 200 and view data 111
1: 3.7
Key word view data 200 and view data 111
2: 66.3
Key word view data 200 and view data 111
3: 31.7
Like this, the order to the similarity of key word view data 200 (shape facility amount) is: first is view data 111
1(with reference to Fig. 8 (b)), second is view data 111
3(with reference to Fig. 8 (d)), the 3rd is view data 111
2(with reference to Fig. 8 (c)).
In this case, search part 106 shows result for retrieval picture 320 shown in Figure 180 on display part 102.In this result for retrieval picture 320, shown the result for retrieval under the situation of utilizing the shape facility amount, this figure from left to right shown key word view data 200, by the view data 111 of the height order of similarity
1, view data 111
3And view data 111
2
Simultaneously, in order to understand,, will feel dissimilar view data 111 to the people corresponding to key word view data 200 (picture with scenes) from result for retrieval picture 320
1Be presented on the high position of similarity, will have felt more similar view data 111
3Being presented at similarity is on the position intermediate.
As mentioned above, according to present embodiment, because type (color configuration characteristic quantity for each characteristic quantity of from key word view data 200, extracting out, texture characteristic amount, the shape facility amount), feel that whether calculating corresponding to similarity is as having or not adaptability to judge to the similar of people, according to this judged result, calculate the similarity of each view data of key word view data 200 and searching object, the view data corresponding with similarity as result for retrieval (with reference to Figure 16～Figure 18) export, thereby can prevent from the selection mistake of the type of applied characteristic quantity when calculating similarity to improve retrieval precision.
In addition, according to present embodiment, in step SC5 shown in Figure 6, has the type of characteristic quantity of adaptability and each view data of searching object has or not adaptability to judge to being judged as for key word image 200, for each view data, get rid of from searching object being judged as each view data that does not have adaptability, each view data that is judged as adaptability is gathered as searching object, the characteristic quantity of type of service calculates the similarity of each view data of key word view data 200 and this searching object, thereby the set by searching object, can further improve retrieval precision and recall precision.
More than, in conjunction with the accompanying drawings first embodiment of the present invention is had been described in detail, but concrete structure example is not limited to this embodiment, in the scope that does not break away from technological thought of the present invention,, also belong to category of the present invention even design alteration etc. is arranged.
For example, in described in front first embodiment, also can realize that the functional programs of image retrieving apparatus shown in Figure 1 100 is stored in the storage medium 500 of embodied on computer readable shown in Figure 19 with being used to, also carry out by making computing machine 400 read these storage medium 500 stored programs, realize the function of image retrieving apparatus 100.
Computing machine 400 by the input media 420 of CPU (central processing unit) 410, keyboard and the mouse etc. of carrying out said procedure, store various data ROM (ROM (read-only memory)) 430, store computing parameter etc. RAM (random access memory) 440, constituted from the output unit 460 of reading device 450, display and the printer etc. of 500 fetch programs of storage medium, the bus 470 of each one of coupling arrangement.
CPU410 by executive routine, realizes the function of image retrieving apparatus 100 read the stored program of storage medium 500 via reading device 450 after.In addition, as storage medium 500, can enumerate CD, floppy disk, hard disk etc.
In addition, in the first embodiment, also can constitute like this: exist under a plurality of situations being judged as type (color configuration characteristic quantity, texture characteristic amount etc.) by adaptability judging part 104 with adaptability, in similarity calculating part 105, the result who the similarity that each type is calculated is carried out integration is as the calculation of similarity degree result.
Constitute according to this, exist under a plurality of situations being judged as type,, thereby, can improve retrieval precision from comprehensive viewpoint because the result that the similarity that each type is calculated is carried out integration is as the calculation of similarity degree result with adaptability.
In addition, in the first embodiment, also can constitute like this: be judged as type (color configuration characteristic quantity, texture characteristic amount etc.) at adaptability judging part 104 and exist under a plurality of situations with adaptability, allow the user in the middle of a plurality of types, select a type, use the characteristic quantity of user-selected type, in similarity calculating part 105, calculate the similarity of each view data of key word view data 200 and searching object.
In addition, in the first embodiment, allow the user confirm the judged result of the adaptability of adaptability judging part 104,, in similarity calculating part 105, calculate the similarity of each view data of key word view data 200 and searching object according to obtaining the judged result that the user confirms.
According to this kind formation, can provide the relevant support of selection with the type of optimal characteristic quantity to the user, can make the user interface of image retrieval better.
As mentioned above, according to the present invention, type for each characteristic quantity of from key word image 200, extracting out, feel that whether having calculated corresponding to similarity is as having or not adaptability to judge to the similar of people, according to this judged result, calculate the similarity of each image of key word image and searching object, exporting as result for retrieval with the corresponding image of similarity, thereby can prevent from the selection mistake of the type of employed characteristic quantity when calculating similarity from can reach the effect that improves retrieval precision.
In addition, according to the present invention, has the type of characteristic quantity of adaptability and each view data of searching object has or not adaptability to judge to being judged as for key word image 200, for each view data, get rid of from searching object being judged as each view data that does not have adaptability, each image that is judged as adaptability is gathered as searching object, the characteristic quantity of type of service calculates the similarity of each view data of key word image and this searching object, thereby the set by searching object, can reach the effect of further raising retrieval precision and recall precision.
In addition, according to the present invention, exist under a plurality of situations being judged as type,, thereby can reach the effect that can improve retrieval precision from comprehensive viewpoint because the result that the similarity that each type is calculated is carried out integration is as the calculation of similarity degree result with adaptability.
In addition, according to the present invention, exist under a plurality of situations being judged as type with adaptability, owing to allow the user in the middle of a plurality of types, select a type, use the characteristic quantity of selected type to calculate the similarity of each image of key word image and searching object, thereby can provide the relevant support of selection with the type of optimal characteristic quantity to the user, can reach the better effect of the user interface that makes image retrieval.
In addition, according to the present invention, owing to allow the user confirm the judged result of adaptability, calculate the similarity of each image of key word image and searching object according to the judged result that has obtained confirming, thereby can provide the relevant support of selection with the type of optimal characteristic quantity to the user, can reach the better effect of the user interface that makes image retrieval.
As mentioned above, image retrieval program of the present invention is useful to the image retrieval of the characteristic quantity data of using a plurality of types (for example, color configuration characteristic quantity, texture characteristic amount, shape facility amount).
Claims (5)
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/JP2002/012135 WO2004047026A1 (en) | 2002-11-20 | 2002-11-20 | Image search program |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1685365A true CN1685365A (en) | 2005-10-19 |
| CN100346339C CN100346339C (en) | 2007-10-31 |
Family
ID=32321512
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB02829713XA Expired - Fee Related CN100346339C (en) | 2002-11-20 | 2002-11-20 | Image retrieval method and image retrieval device |
Country Status (3)
| Country | Link |
|---|---|
| JP (1) | JP4160050B2 (en) |
| CN (1) | CN100346339C (en) |
| WO (1) | WO2004047026A1 (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101902578A (en) * | 2009-05-26 | 2010-12-01 | 三洋电机株式会社 | Image-reproducing means and filming apparatus |
| CN101266612B (en) * | 2007-03-13 | 2011-08-17 | 株式会社理光 | Image retrieval apparatus, image retrieving method |
| CN101896901B (en) * | 2007-12-12 | 2013-06-19 | 微软公司 | Interactive concept learning in image search |
| CN101911077B (en) * | 2007-12-27 | 2016-05-11 | 皇家飞利浦电子股份有限公司 | For the method and apparatus of hierarchical search |
| CN105959561A (en) * | 2016-06-13 | 2016-09-21 | Tcl集团股份有限公司 | Photographing method and system |
| CN107798325A (en) * | 2017-08-18 | 2018-03-13 | 中国银联股份有限公司 | Card identification method and equipment, computer-readable storage medium |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4556121B2 (en) * | 2005-01-07 | 2010-10-06 | ソニー株式会社 | Information processing apparatus and method, and program |
| JP4792576B2 (en) * | 2005-04-07 | 2011-10-12 | 国立大学法人広島大学 | Data storage device, data retrieval device, data reproduction device, and data retrieval program |
| JP4859025B2 (en) * | 2005-12-16 | 2012-01-18 | 株式会社リコー | Similar image search device, similar image search processing method, program, and information recording medium |
| JP4770863B2 (en) * | 2008-04-08 | 2011-09-14 | 富士ゼロックス株式会社 | Image processing apparatus and image processing program |
| US20100166303A1 (en) * | 2008-12-31 | 2010-07-01 | Ali Rahimi | Object recognition using global similarity-based classifier |
| JP4906900B2 (en) * | 2009-09-24 | 2012-03-28 | ヤフー株式会社 | Image search apparatus, image search method and program |
| JP5668932B2 (en) * | 2011-05-23 | 2015-02-12 | 株式会社モルフォ | Image identification device, image identification method, image identification program, and recording medium |
| CN103310221B (en) * | 2012-03-16 | 2016-04-13 | 富士通株式会社 | Image processing apparatus, image processing method and equipment |
| JP6091217B2 (en) * | 2013-01-08 | 2017-03-08 | キヤノン株式会社 | Image search device, image search method, search source image providing device, search source image providing method, and program |
| JP2020095757A (en) * | 2020-03-23 | 2020-06-18 | キヤノン株式会社 | Information processing device, information processing method, and program |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5802361A (en) * | 1994-09-30 | 1998-09-01 | Apple Computer, Inc. | Method and system for searching graphic images and videos |
| US6072904A (en) * | 1997-12-31 | 2000-06-06 | Philips Electronics North America Corp. | Fast image retrieval using multi-scale edge representation of images |
| JP2000187731A (en) * | 1998-12-21 | 2000-07-04 | Ricoh Co Ltd | Picture feature extraction method and recording medium which records program for making computer execute respective processes of the same and which computer can read |
| KR100319151B1 (en) * | 1999-02-01 | 2002-01-05 | 구자홍 | Data structure of multi-level image and method for image checking using this structure |
| JP3978962B2 (en) * | 2000-01-13 | 2007-09-19 | オムロン株式会社 | Information search method and information search apparatus |
| KR100677096B1 (en) * | 2000-05-31 | 2007-02-05 | 삼성전자주식회사 | Image similarity evaluation method and apparatus |
| KR100788643B1 (en) * | 2001-01-09 | 2007-12-26 | 삼성전자주식회사 | Image retrieval method based on combination of color and texture |
-
2002
- 2002-11-20 CN CNB02829713XA patent/CN100346339C/en not_active Expired - Fee Related
- 2002-11-20 WO PCT/JP2002/012135 patent/WO2004047026A1/en active Application Filing
- 2002-11-20 JP JP2004553119A patent/JP4160050B2/en not_active Expired - Fee Related
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101266612B (en) * | 2007-03-13 | 2011-08-17 | 株式会社理光 | Image retrieval apparatus, image retrieving method |
| CN101896901B (en) * | 2007-12-12 | 2013-06-19 | 微软公司 | Interactive concept learning in image search |
| CN101911077B (en) * | 2007-12-27 | 2016-05-11 | 皇家飞利浦电子股份有限公司 | For the method and apparatus of hierarchical search |
| CN101902578A (en) * | 2009-05-26 | 2010-12-01 | 三洋电机株式会社 | Image-reproducing means and filming apparatus |
| CN105959561A (en) * | 2016-06-13 | 2016-09-21 | Tcl集团股份有限公司 | Photographing method and system |
| CN107798325A (en) * | 2017-08-18 | 2018-03-13 | 中国银联股份有限公司 | Card identification method and equipment, computer-readable storage medium |
| CN107798325B (en) * | 2017-08-18 | 2021-04-16 | 中国银联股份有限公司 | Card recognition method and apparatus, computer storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| JP4160050B2 (en) | 2008-10-01 |
| JPWO2004047026A1 (en) | 2006-03-16 |
| WO2004047026A1 (en) | 2004-06-03 |
| CN100346339C (en) | 2007-10-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100346339C (en) | Image retrieval method and image retrieval device | |
| CN1573742A (en) | Image retrieving system, image classifying system, image retrieving program, image classifying program, image retrieving method and image classifying method | |
| CN1669052A (en) | Image matching system using 3-dimensional object model, image matching method, and image matching program | |
| CN1175368C (en) | Three-dimensional moulding device and method | |
| CN1184796C (en) | Image processing method and equipment, image processing system and storage medium | |
| CN1218265C (en) | Device and method for database searching | |
| CN1551017A (en) | Image searching device, method, programmme and storage medium storing said programme | |
| CN1790376A (en) | Information processing apparatus , input device and method, program and information processing system | |
| CN1519757A (en) | Image retrieval device, keyword assignment method and program for image retrieval device | |
| CN1925541A (en) | Image processing apparatus and method thereof | |
| CN1854946A (en) | Manufacture condition setting system and method | |
| CN1685364A (en) | Method for segmenting pixels in an image | |
| CN101069184A (en) | Information processing device, method, and program | |
| CN1917578A (en) | Data processing apparatus,data processing method and program | |
| CN1496093A (en) | Image information distributing method and system, central and terminal equipment and scanistor | |
| CN1871563A (en) | Machining information creating device, program, and machining information creating method | |
| CN1696959A (en) | Specific subject detection device | |
| CN1920825A (en) | Method and system for displaying performance constraints in a flow design tool | |
| CN1940913A (en) | Variable table data type setting method | |
| CN1588431A (en) | Character extracting method from complecate background color image based on run-length adjacent map | |
| CN1577350A (en) | Information display apparatus and method | |
| CN1227612C (en) | Color quantization method and multimedia search method using the method | |
| CN1942899A (en) | Face image creating device and method | |
| CN1277398A (en) | Method and apparatus for document retrieval | |
| CN1254955C (en) | Image displaying device and controlling method, camera therewith |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20071031 Termination date: 20151120 |
|
| EXPY | Termination of patent right or utility model |