KR20160127369A - System and method for searching image - Google Patents
System and method for searching image Download PDFInfo
- Publication number
- KR20160127369A KR20160127369A KR1020150058655A KR20150058655A KR20160127369A KR 20160127369 A KR20160127369 A KR 20160127369A KR 1020150058655 A KR1020150058655 A KR 1020150058655A KR 20150058655 A KR20150058655 A KR 20150058655A KR 20160127369 A KR20160127369 A KR 20160127369A
- Authority
- KR
- South Korea
- Prior art keywords
- image
- dimensional
- index value
- index
- depth
- Prior art date
Links
Images
Classifications
-
- G06F17/30244—
-
- G06F17/30256—
-
- G06F17/30277—
-
- G06F17/3028—
-
- G06T7/0028—
Abstract
An image retrieval system and method are provided. According to an embodiment of the present invention, there is provided an image retrieval system including: a depth image obtaining unit that obtains a two-dimensional depth image representing a contour of the three-dimensional image from a target file including a three-dimensional image; An index generator for extracting a feature value from the 2D depth image and generating an index value for the 3D image from the feature value; And an image retrieving unit for retrieving an index value having the highest similarity to the generated index value among the stored index values and retrieving a three-dimensional image corresponding to the index value extracted from the stored three-dimensional images.
Description
Embodiments of the present invention relate to a technique capable of retrieving a three-dimensional image at high speed.
In recent years, 3D printing technology is gradually being commercialized. Accordingly, efforts are being made to search a large-capacity three-dimensional image file at high speed in order to prevent unauthorized copying, alteration and abuse of a three-dimensional image file in advance.
However, since the conventional three-dimensional image retrieval technology computes a three-dimensional image file in a three-dimensional coordinate system, the complexity of the computation becomes three-quadrants of the data amount, There is a problem that image retrieval speed is remarkably slow. In order to solve these problems, a retrieval method of inputting words (keywords), sentences, categories or the like describing the appearance of a three-dimensional image or using metadata of a three-dimensional image file has been suggested as an alternative. However, There was a very low problem. In addition, a compound technique for simultaneously processing words (keywords), sentences, categories, and metadata describing the outline of a three-dimensional image together with external data of a three-dimensional image has been proposed. However, In the data age, it is difficult to efficiently retrieve large amounts of data.
Embodiments of the present invention are intended to provide a means for efficiently searching a large-capacity three-dimensional image file.
According to an exemplary embodiment of the present invention, there is provided an image processing apparatus including: a depth image obtaining unit obtaining a two-dimensional depth image representing a contour of the three-dimensional image from a target file including a three-dimensional image; An index generator for extracting a feature value from the 2D depth image and generating an index value for the 3D image from the feature value; And an image retrieval unit for extracting an index value having the highest similarity to the generated index value among the stored index values and retrieving a three-dimensional image corresponding to the index value extracted from the stored three- / RTI >
The depth image obtaining unit may obtain a plurality of the two-dimensional depth images by mapping the pixels of the three-dimensional image to the two-dimensional plane in multiple directions based on the set axis.
The image search system may further include a fingerprint image obtaining unit for normalizing the two-dimensional depth image to obtain a two-dimensional fingerprint image, and the index generating unit may extract the feature value from the two-dimensional fingerprint image.
The fingerprint image obtaining unit may obtain the two-dimensional fingerprint image by latticing the depth image into a plurality of cells and normalizing the number of pixels in the cell and the thickness of the outline of the depth image.
The index generator may calculate the feature value by multiplying the sum of the normalized number of pixels included in the two-dimensional fingerprint image and the variance value of the number of normalized pixels included in each cell of the two-dimensional fingerprint image with respect to the sum have.
Wherein the fingerprint image obtaining unit obtains a plurality of the two-dimensional fingerprint images for the plurality of depth images, and obtains a two-dimensional fingerprint image having the largest feature value among the plurality of obtained two- Can be selected.
The index generator may select variance values of the normalized number of pixels included in each cell of the representative two-dimensional fingerprint image as an index value for the three-dimensional image.
The image searching unit may calculate a difference between the stored index value and the generated index value and extract an index value having the highest similarity to the generated index value among the stored index values.
According to another exemplary embodiment of the present invention, there is provided a depth image obtaining unit, comprising: obtaining a two-dimensional depth image representing a contour of the three-dimensional image from a target file including a three-dimensional image; Extracting a feature value from the 2D depth image; Generating an index value for the three-dimensional image from the feature value; Extracting an index value having the highest degree of similarity with the generated index value among the stored index values in the image searching unit; And retrieving, in the image retrieving unit, a three-dimensional image corresponding to the index value extracted from the stored three-dimensional image.
The acquiring of the two-dimensional depth image may acquire a plurality of the two-dimensional depth images by mapping the pixels of the three-dimensional image to the two-dimensional plane in multiple directions based on the set axis.
Wherein the image retrieval method further comprises the step of obtaining a two-dimensional fingerprint image by normalizing the two-dimensional depth image in a fingerprint image obtaining unit after obtaining the two-dimensional depth image, The feature value may be extracted from the two-dimensional fingerprint image.
The step of acquiring the two-dimensional fingerprint image may include gridding the depth image into a plurality of cells; And normalizing the thickness of the contour of the depth image with the number of pixels in the cell.
Wherein the extracting of the feature value comprises multiplying the sum of the normalized number of pixels included in the two-dimensional fingerprint image and the variance value of the number of normalized pixels included in each cell of the two-dimensional fingerprint image with respect to the sum, The value can be calculated.
Wherein acquiring the two-dimensional fingerprint image comprises: obtaining a plurality of the two-dimensional fingerprint images for the plurality of depth images; And selecting a two-dimensional fingerprint image having the largest feature value among the acquired plurality of two-dimensional fingerprint images as a representative two-dimensional fingerprint image.
The step of generating an index value for the three-dimensional image may include, as index values for the three-dimensional image, variance values of the number of normalized pixels contained in each cell of the representative two-dimensional fingerprint image.
The step of extracting the index value having the highest degree of similarity may include calculating a difference between the stored index value and the generated index value and extracting an index value having the highest degree of similarity with the generated index value .
According to another exemplary embodiment of the present invention, there is provided an image processing method comprising the steps of: obtaining, in a depth image acquiring unit, a two-dimensional depth image representing a contour of a three-dimensional image from a three-dimensional image; Extracting a feature value from the 2D depth image; Generating an index value for the three-dimensional image from the feature value; Extracting an index value having the highest degree of similarity with the generated index value among the stored index values in the image searching unit; And retrieving, in the image retrieving unit, a three-dimensional image corresponding to the index value extracted from the stored three-dimensional image.
According to embodiments of the present invention, a three-dimensional image is converted into a one-dimensional index value, and a stored three-dimensional image is retrieved using the converted one-dimensional index value, so that the complexity of an operation required for image retrieval is significantly . As a result, the overall image search speed can be improved and the image search time can be shortened. In addition, since the index value is generated based on the outline of the three-dimensional image, the accuracy of the image search is high.
1 is a block diagram showing a detailed configuration of an image retrieval system according to an embodiment of the present invention;
2 is a view for explaining a process of acquiring a 2D depth image by a depth image acquiring unit according to an embodiment of the present invention;
3 is a view for explaining a process of acquiring a two-dimensional fingerprint image according to an embodiment of the present invention;
FIG. 4 is a view for explaining a process of extracting a feature value by an index generating unit according to an embodiment of the present invention.
5 is a view for explaining a process of measuring the similarity between an index value stored in an image search unit and an index value generated in an index generator according to an embodiment of the present invention;
6 is a flowchart illustrating an image search method according to an embodiment of the present invention.
Hereinafter, specific embodiments of the present invention will be described with reference to the drawings. The following detailed description is provided to provide a comprehensive understanding of the methods, apparatus, and / or systems described herein. However, this is merely an example and the present invention is not limited thereto.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description, well-known functions or constructions are not described in detail since they would obscure the invention in unnecessary detail. The following terms are defined in consideration of the functions of the present invention, and may be changed according to the intention or custom of the user, the operator, and the like. Therefore, the definition should be based on the contents throughout this specification. The terms used in the detailed description are intended only to describe embodiments of the invention and should in no way be limiting. Unless specifically stated otherwise, the singular form of a term includes plural forms of meaning. In this description, the expressions "comprising" or "comprising" are intended to indicate certain features, numbers, steps, operations, elements, parts or combinations thereof, Should not be construed to preclude the presence or possibility of other features, numbers, steps, operations, elements, portions or combinations thereof.
1 is a block diagram showing a detailed configuration of an
The depth
The depth
The fingerprint
The
The
Also, the
The
In one embodiment, the depth
2 is a view for explaining a process of acquiring a 2D depth image by the depth
FIG. 3 is a view for explaining a process of acquiring a two-dimensional fingerprint image by the fingerprint
4 is a diagram for explaining a process of extracting a feature value by the
As described above, the fingerprint
The
5 is a diagram for explaining a process of measuring the similarity between the index value stored in the
Here, d is the difference between the index value Qi stored in the
According to embodiments of the present invention, a three-dimensional image is converted into a one-dimensional index value, and a stored three-dimensional image is retrieved using the converted one-dimensional index value, so that the complexity of an operation required for image retrieval is significantly . As a result, the overall image search speed can be improved and the image search time can be shortened. In addition, since the index value is generated based on the outline of the three-dimensional image, the accuracy of the image search is high.
6 is a flowchart illustrating an image search method according to an embodiment of the present invention. The method shown in Fig. 6 can be performed, for example, by the
In step S602, the depth
In step S604, the fingerprint
In step S606, the
In step S608, the fingerprint
In step S610, the
In step S612, the
On the other hand, an embodiment of the present invention may include a program for performing the methods described herein on a computer, and a computer-readable recording medium including the program. The computer-readable recording medium may include a program command, a local data file, a local data structure, or the like, alone or in combination. The media may be those specially designed and constructed for the present invention, or may be those that are commonly used in the field of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape, optical recording media such as CD-ROMs and DVDs, and specifically configured to store and execute program instructions such as ROM, RAM, flash memory, Hardware devices. Examples of such programs may include machine language code such as those produced by a compiler, as well as high-level language code that may be executed by a computer using an interpreter or the like.
While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, I will understand. Therefore, the scope of the present invention should not be limited to the above-described embodiments, but should be determined by equivalents to the appended claims, as well as the appended claims.
100: Image Search System
102: Depth image generation unit
104: fingerprint image generating unit
106: index generation unit
108: Image search unit
110: Database
Claims (17)
An index generator for extracting a feature value from the 2D depth image and generating an index value for the 3D image from the feature value; And
And an image retrieval unit for retrieving an index value having the highest similarity to the generated index value among the stored index values and searching for a three-dimensional image corresponding to the index value extracted from the stored three-dimensional images.
Wherein the depth image obtaining unit obtains a plurality of the two-dimensional depth images by mapping pixels of the three-dimensional image to a two-dimensional plane in multiple directions based on the set axis.
Further comprising a fingerprint image acquiring unit for acquiring a two-dimensional fingerprint image by normalizing the two-dimensional depth image,
Wherein the index generating unit extracts the feature value from the two-dimensional fingerprint image.
Wherein the fingerprint image obtaining unit obtains the two-dimensional fingerprint image by latticing the depth image into a plurality of cells, and normalizing the number of pixels in the cell and the thickness of the contour of the depth image.
Wherein the index generator calculates the feature value by multiplying the sum of the normalized number of pixels included in the two-dimensional fingerprint image by the variance of the number of normalized pixels included in each cell of the two-dimensional fingerprint image with respect to the sum , Image retrieval system.
Wherein the fingerprint image obtaining unit obtains a plurality of the two-dimensional fingerprint images for the plurality of depth images, and obtains a two-dimensional fingerprint image having the largest feature value among the plurality of obtained two- Image search system to select.
Wherein the index generating unit selects the variance values of the normalized number of pixels included in each cell of the representative two-dimensional fingerprint image as an index value for the three-dimensional image.
Wherein the image searching unit calculates a difference between the stored index value and the generated index value and extracts an index value having the highest degree of similarity with the generated index value among the stored index values.
Extracting a feature value from the 2D depth image;
Generating an index value for the three-dimensional image from the feature value;
Extracting an index value having the highest degree of similarity with the generated index value among the stored index values in the image searching unit; And
And searching the image searching unit for a three-dimensional image corresponding to the index value extracted from the stored three-dimensional image.
Wherein the acquiring of the two-dimensional depth image acquires a plurality of the two-dimensional depth images by mapping the pixels of the three-dimensional image to the two-dimensional plane in multiple directions based on the set axis.
After obtaining the two-dimensional depth image,
In the fingerprint image acquisition unit, further comprising the step of normalizing the two-dimensional depth image to obtain a two-dimensional fingerprint image,
Wherein the extracting of the feature value extracts the feature value from the two-dimensional fingerprint image.
Wherein acquiring the two-dimensional fingerprint image comprises:
Grating the depth image into a plurality of cells; And
Normalizing the number of pixels in the cell and the thickness of the contour of the depth image.
Wherein the extracting of the feature value comprises multiplying the sum of the normalized number of pixels included in the two-dimensional fingerprint image and the variance value of the number of normalized pixels included in each cell of the two-dimensional fingerprint image with respect to the sum, A method of image retrieval that computes a value.
Wherein acquiring the two-dimensional fingerprint image comprises:
Obtaining a plurality of the two-dimensional fingerprint images for the plurality of depth images; And
Selecting a representative two-dimensional fingerprint image as a two-dimensional fingerprint image having the largest feature value among the acquired plurality of two-dimensional fingerprint images.
Wherein the step of generating an index value for the three-dimensional image comprises selecting the variance values of the normalized number of pixels included in each cell of the representative two-dimensional fingerprint image as an index value for the three-dimensional image.
The step of extracting the index value having the highest degree of similarity may include calculating a difference between the stored index value and the generated index value and extracting an index value having the highest degree of similarity with the generated index value among the stored index values , Image retrieval method.
In the depth image acquiring section, acquiring a depth image representing a contour of the three-dimensional image from the three-dimensional image;
Extracting a feature value from the 2D depth image;
Generating an index value for the three-dimensional image from the feature value;
Extracting an index value having the highest degree of similarity with the generated index value among the stored index values in the image searching unit; And
The image searching unit searches for a three-dimensional image corresponding to the index value extracted from the stored three-dimensional images
The computer program being stored on a recording medium.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020150058655A KR20160127369A (en) | 2015-04-27 | 2015-04-27 | System and method for searching image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020150058655A KR20160127369A (en) | 2015-04-27 | 2015-04-27 | System and method for searching image |
Publications (1)
Publication Number | Publication Date |
---|---|
KR20160127369A true KR20160127369A (en) | 2016-11-04 |
Family
ID=57530314
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
KR1020150058655A KR20160127369A (en) | 2015-04-27 | 2015-04-27 | System and method for searching image |
Country Status (1)
Country | Link |
---|---|
KR (1) | KR20160127369A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110390030A (en) * | 2019-06-28 | 2019-10-29 | 中山大学 | The storage method and device of pictorial information |
KR20190123842A (en) * | 2018-04-25 | 2019-11-04 | 광주과학기술원 | Operating method of a system for reconstucting 3-d shapes using neural network |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20140004240A (en) | 2011-09-29 | 2014-01-10 | 텐센트 테크놀로지(센젠) 컴퍼니 리미티드 | Image browsing method, system and computer storage medium |
-
2015
- 2015-04-27 KR KR1020150058655A patent/KR20160127369A/en unknown
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20140004240A (en) | 2011-09-29 | 2014-01-10 | 텐센트 테크놀로지(센젠) 컴퍼니 리미티드 | Image browsing method, system and computer storage medium |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20190123842A (en) * | 2018-04-25 | 2019-11-04 | 광주과학기술원 | Operating method of a system for reconstucting 3-d shapes using neural network |
CN110390030A (en) * | 2019-06-28 | 2019-10-29 | 中山大学 | The storage method and device of pictorial information |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9280827B2 (en) | Method for determining object poses using weighted features | |
CN110413816B (en) | Color Sketch Image Search | |
US10311099B2 (en) | Method and system for 3D model database retrieval | |
US9165217B2 (en) | Techniques for ground-level photo geolocation using digital elevation | |
US20150278155A1 (en) | Identifying objects using a 3d scanning device, images, and 3d models | |
US20160027208A1 (en) | Image analysis method | |
US11475593B2 (en) | Methods and apparatus for processing image data for machine vision | |
AU2018202767B2 (en) | Data structure and algorithm for tag less search and svg retrieval | |
US20240096123A1 (en) | Methods and apparatus for testing multiple fields for machine vision | |
WO2011105134A1 (en) | Position and orientation estimation method and apparatus therefor | |
US10846563B2 (en) | Methods and apparatus for generating a dense field of three dimensional data for machine vision | |
Wang et al. | From low-cost depth sensors to cad: Cross-domain 3d shape retrieval via regression tree fields | |
KR20160127369A (en) | System and method for searching image | |
JP4570995B2 (en) | MATCHING METHOD, MATCHING DEVICE, AND PROGRAM | |
KR20240013085A (en) | Methods and apparatus for processing image data for machine vision | |
Manuel et al. | A hybrid approach for the semantic annotation of spatially oriented images | |
KR20190070680A (en) | Apparatus and method for retrieving and repairing part for maintaining partial breakage of part, and 3d printing based part maintenance system | |
JP2018136642A (en) | Three-dimensional shape search method and three-dimensional shape search system | |
KR101692634B1 (en) | Method and Device for Transforming 2D Image into 3D | |
Lu et al. | Three-dimensional object recognition using an extensible local surface descriptor | |
KR102068489B1 (en) | 3d object creation apparatus | |
CN113838005B (en) | Intelligent identification and three-dimensional reconstruction method and system for rock mass fracture based on dimension conversion | |
Drzewiecki et al. | Comparison of selected textural features as global content-based descriptors of VHR satellite image-the EROS-a study | |
Lu et al. | Simple and efficient improvement of spin image for three-dimensional object recognition | |
KR102054211B1 (en) | Method and system for video retrieval based on image queries |