CN105069026B - A kind of Chinese character image lookup method based on optical flow analysis - Google Patents
A kind of Chinese character image lookup method based on optical flow analysis Download PDFInfo
- Publication number
- CN105069026B CN105069026B CN201510422929.8A CN201510422929A CN105069026B CN 105069026 B CN105069026 B CN 105069026B CN 201510422929 A CN201510422929 A CN 201510422929A CN 105069026 B CN105069026 B CN 105069026B
- Authority
- CN
- China
- Prior art keywords
- optical flow
- chinese character
- image
- character image
- flow analysis
- 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.)
- Expired - Fee Related
Links
- 230000003287 optical effect Effects 0.000 title claims abstract description 22
- 238000005206 flow analysis Methods 0.000 title claims abstract description 13
- 238000000034 method Methods 0.000 title claims abstract description 11
- 238000010276 construction Methods 0.000 claims description 5
- 238000006073 displacement reaction Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 abstract description 2
- 238000001914 filtration Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
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
- G06F16/58—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
Landscapes
- Engineering & Computer Science (AREA)
- Library & Information Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Character Input (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of Chinese character image lookup method based on optical flow analysis.Chinese character database in this method is directly stored with image, and user's input is also Chinese character image when lookup.User inputs after image to be found, is aligned input picture with every image in Chinese character image library first with optical flow analysis, then the distance between measurement.Several the smallest images of lookup result selected distance.
Description
Technical field
The invention belongs to computer vision and field of image processing, specifically optical flow analysis and image lookup method,
The Chinese character image that this method can input user, is aligned as much as possible with the Chinese character image in library using optical flow analysis, is then spent
Input picture is measured at a distance from image in library, chooses the image of several minimum distances as output result.
Background technique
With stepping up for Chinese national power, Chinese character cultural relics and historic sites are increasingly taken seriously, and utilize digital image acquisition skill
Art, a large amount of Chinese character an inscription on a tablet image are stored in database.Carry out Chinese character image lookup, in terms of work also by
One expansion.Traditional looking up Chinese characters mode is manual identified to be carried out by some ancient Chinese prose experts and in the form of a label to each Chinese
Word image is stored, and lookup when is also to search for database in a manner of searching label, this needs a large amount of artificial ginseng
With.And the mode of handmarking is easy to be influenced by subjective judgement.
Summary of the invention
The invention proposes a kind of Chinese character image lookup method based on optical flow analysis.Main inside includes four steps:
The pretreatment and database construction of Chinese character image；The pretreatment of input picture；Image alignment is carried out using optical flow analysis；Measurement pair
The smallest several images of the distance and selected distance of image after neat, which are used as, to be returned the result.
Detailed description of the invention
The pretreatment of Fig. 1 Chinese character image.
Specific embodiment
Chinese character image based on optical flow analysis of the invention searches technology and uses following steps:
(1) pretreatment and database construction of Chinese character image.Each Chinese character is split from the image of the historic sites such as an inscription on a tablet, is used
Binary image indicates (black matrix wrongly written or mispronounced character), fills dark border to the image of non-square and forms square, since image may
There is noise, needs to carry out image certain filtering (such as the deformation of median filtering, morphology), then arrive image down
Key assignments of the small image of 32x32 as each original image, key assignments are used for images match when searching.
(2) pretreatment of input picture.The present invention allows user to input Chinese character and input Chinese character image two ways progress
It searches.If user inputs Chinese character, the Chinese character of input is first transformed into Chinese character image with fontlib (such as lishu font), this
The image that kind mode obtains can directly zoom to 32x32 as key assignments.And the lookup mode for directly inputting Chinese character image,
It is also required to be pre-processed, treatment process is referring to step (1).
(3) optical flow analysis is aligned.The optical flow analysis model that we use is as follows:
E (u, v)=Edata(u,v)+α2Esm(u,v) (1)
(u, v) is optical flow field to be asked in formula, and u is horizontal displacement components, and v is vertical displacement component, and unit is pixel,
EdataFor Data Matching item energy, EsmFor flatness energy, α is weighting coefficient.Wherein:
Edata(u, v)=∫Ω||I1(x,y)-I2(x+u,y+v)||1dΩ (2)
I1, I2Respectively two matched images, Ω are integral space, and (x, y) is space coordinate, Data Matching energy term
We use a norm, this allows for the Chinese character image in database and has noise, more efficient when using a norm.
When solving this optical flow estimation, we construct image pyramid first, and progress solves from thick to thin, pyramid construction
3 grades are respectively 8x8, and 16x16,32x32. initial optical flow field is initialized as 0, when the optical flow field of low resolution solves completion
Afterwards, optical flow field high level resolution ratio is upsampled to as initial value to solve.And the solution procedure of each level resolution ratio
It is as follows:
(a) formula is usedInstead of the norm in formula (2).
(b) Euler-Lagrange equation (the namely equation of optimal solution satisfaction) of formula (1) are as follows:
Wherein:
It=I2(x+u,y+v)-I1(x,y)
Due to formula (3) (4) be it is nonlinear, we are solved using two layers of stable point alternative manner；
Outer layer stable point iteration are as follows:
It is rightCarry out first order Taylor expansion:
Wherein uk+1=uk+duk, vk+1=vk+dvkVariable is by (uk+1,vk+1) become (duk,dvk),
Corresponding formula (5), (6) become:
Then internal layer stable point iteration is carried out:
(4) measure image distance with return the result.After finding out optical flow field (u, v) in step (3), by image I2Alignment
To image I1Then metric range, distance definition are as follows:
D(I1,I2)=∫Ω||I1(x,y)-I2(x+u,y+v)||1dΩ。
Claims (1)
1. a kind of Chinese character image lookup method based on optical flow analysis, comprising the following steps:
(1) pretreatment and database construction of Chinese character image,
(2) pretreatment of input picture,
(3) optical flow analysis is aligned；
In the step (3), the data capacity item of the optical flow estimation of construction samples L1 norm, and smooth item uses L2 norm；
(4) it measures image and returns the result；
In the step (4), metric range has used the mode of the sum of absolute value subtracted each other pixel-by-pixel；
Optical flow estimation in the step (3) is as follows:
E (u, v)=Edata(u,v)+α2Esm(u,v)
(u, v) is optical flow field to be asked in formula, and u is horizontal displacement components, and v is vertical displacement component, and unit is pixel, EdataFor
Data Matching item energy, EsmFor flatness energy, α is weighting coefficient；Wherein:
Edata(u, v)=∫Ω||I1(x,y)-I2(x+u,y+v)||1dΩ
I1, I2Respectively two matched images, Ω are integral space, and (x, y) is space coordinate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510422929.8A CN105069026B (en) | 2015-07-17 | 2015-07-17 | A kind of Chinese character image lookup method based on optical flow analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510422929.8A CN105069026B (en) | 2015-07-17 | 2015-07-17 | A kind of Chinese character image lookup method based on optical flow analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105069026A CN105069026A (en) | 2015-11-18 |
| CN105069026B true CN105069026B (en) | 2019-03-08 |
Family
ID=54498399
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510422929.8A Expired - Fee Related CN105069026B (en) | 2015-07-17 | 2015-07-17 | A kind of Chinese character image lookup method based on optical flow analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105069026B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103268363A (en) * | 2013-06-06 | 2013-08-28 | 哈尔滨工业大学 | Elastic HOG (histograms of oriented gradient) feature-based Chinese calligraphy image retrieval method matched with DDTW (Derivative dynamic time wrapping) |
| CN104504049A (en) * | 2014-12-20 | 2015-04-08 | 辽宁师范大学 | Retrieval method of color images based on quaternion Harmonic-Fourier moments |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104574276A (en) * | 2015-01-29 | 2015-04-29 | 厦门美图之家科技有限公司 | Method and device for aligning images on basis of optical flow |
-
2015
- 2015-07-17 CN CN201510422929.8A patent/CN105069026B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103268363A (en) * | 2013-06-06 | 2013-08-28 | 哈尔滨工业大学 | Elastic HOG (histograms of oriented gradient) feature-based Chinese calligraphy image retrieval method matched with DDTW (Derivative dynamic time wrapping) |
| CN104504049A (en) * | 2014-12-20 | 2015-04-08 | 辽宁师范大学 | Retrieval method of color images based on quaternion Harmonic-Fourier moments |
Non-Patent Citations (2)
| Title |
|---|
| 基于无人机影像的景象匹配系统设计与实现;赵峰;《中国优秀硕士学位论文全文数据库 信息科技辑》;20130331(第03期);第35-38页 |
| 基于最短欧氏距离匹配的印刷体汉字识别;马飞等;《平顶山学院学报》;20120430;第27卷(第2期);第70-73页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105069026A (en) | 2015-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11238602B2 (en) | Method for estimating high-quality depth maps based on depth prediction and enhancement subnetworks | |
| CN111291885B (en) | Near infrared image generation method, training method and device for generation network | |
| WO2021022929A1 (en) | Single-frame image super-resolution reconstruction method | |
| CN107818554B (en) | Information processing apparatus and information processing method | |
| CN107808131A (en) | Dynamic gesture identification method based on binary channel depth convolutional neural networks | |
| CN103325095B (en) | A kind of openness image repair method of sample block of the bonding position factor | |
| CN113870128B (en) | Digital mural image restoration method based on depth convolution countermeasure network | |
| CN107392880A (en) | A kind of imitative pattern painting automatic generation method | |
| CN112561879B (en) | Ambiguity evaluation model training method, image ambiguity evaluation method and image ambiguity evaluation device | |
| CN106571072A (en) | Method for realizing children education card based on AR | |
| CN111862030B (en) | Face synthetic image detection method and device, electronic equipment and storage medium | |
| CN109284767A (en) | A kind of pedestrian retrieval method based on augmented sample and multithread layer | |
| CN112686830B (en) | Super-resolution method of single depth map based on image decomposition | |
| CN112307847A (en) | Multi-scale attention pedestrian re-recognition deep learning system based on guidance | |
| Wang et al. | A lightweight multi-label segmentation network for mobile iris biometrics | |
| CN111626954B (en) | Mural image color restoration method and device, storage medium and computer equipment | |
| Li et al. | Image reflection removal using end‐to‐end convolutional neural network | |
| Xu et al. | Artistic color virtual reality implementation based on similarity image restoration | |
| CN105069026B (en) | A kind of Chinese character image lookup method based on optical flow analysis | |
| Chen et al. | Dual discriminator gan: Restoring ancient yi characters | |
| Hou et al. | Relative gradients for image lighting correction | |
| Zhang et al. | Research on Semantic Segmentation Based on Improved PSPNet | |
| CN114782249A (en) | Super-resolution reconstruction method, device and equipment for image and storage medium | |
| CN114565777A (en) | Data processing method and device | |
| CN103549960B (en) | Footprint acquisition and extraction device and footprint acquisition and extraction method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190308 |