CN106599028A - Book content searching and matching method based on video image processing - Google Patents
Book content searching and matching method based on video image processing Download PDFInfo
- Publication number
- CN106599028A CN106599028A CN201610946349.3A CN201610946349A CN106599028A CN 106599028 A CN106599028 A CN 106599028A CN 201610946349 A CN201610946349 A CN 201610946349A CN 106599028 A CN106599028 A CN 106599028A
- Authority
- CN
- China
- Prior art keywords
- image
- target
- algorithm
- target image
- book
- 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
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30176—Document
Landscapes
- Engineering & Computer 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)
- Image Analysis (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Processing (AREA)
Abstract
The invention provides a book content searching and matching method based on video image processing. Images captured by a camera are processed, so that searching and matching of book content images in a current camera are carried out. The method comprises the following steps of: performing target image segmentation of the images captured by the camera by using an image processing technology, so that a target image area is obtained; and then, extracting a target image from the target image area through a four-edge detection algorithm, furthermore, coding the target image through a perception hash algorithm, and performing searching and matching in a database according to the code of the target image, so that the page of a current book content is obtained. The intelligent book content searching and matching method provided by the invention is particularly suitable for education products, such as children's robots, and has wide market prospect and practical significance.
Description
Technical field
The present invention relates to be based on image/video cutting techniques core Hash code search matching technique, and in particular to based on regarding
The book contents searching and matching method of frequency image procossing.
Background technology
In recent years, the robot teaching of China has a great development, machine under the promotion energetically that national computer is educated
Device people has become the very high creative teaching platform of cry.Children robot in the market is mainly child and provides company work(
Can, more such as about education in terms of "smart" products, mainly have talking pen, point reader, children's panel computer etc., these religion
Although it is more ripe to educate product technology, itself technical difficulty is not high, and outfit of equipment function is simpler, and educational function is then
Relatively fewer or mistake is single, it is impossible to provides the complicated problems in teaching of a bit, therefore just very provides learning guide for children.
In addition, have the mobile learning platform such as " monarch scholar-tyrant, one who exercises autocratic control in academic and educational circles " and " operation side " that study can be provided on market answer, but not
Pipe is that " monarch scholar-tyrant, one who exercises autocratic control in academic and educational circles " or " operation side " is required for being taken pictures on textbook in use, then benefits from adjustment and intercepts
Frame intercepts problem area, and this has some troubles in interaction.Therefore, for children education, in the urgent need to a kind of energy on market
The scheme of enough intelligently guidance children for learning.
Therefore, the invention provides a kind of book contents searching and matching method based on Computer Vision, by religion
Educate and install in robot camera, camera shoots to book contents when child does one's assignment, ask when child runs into
During topic, Video Image Segmentation and search matching feature are opened, the image to photographing is processed, current so as to obtain child
The content for learning only needs in addition child and says " which topic " in conjunction with speech recognition in the page number of correspondence books
Exercise question number is obtained, so as to obtaining exercise question that child currently doing and can furnishing an answer from backstage.
The content of the invention
It is an object of the invention to introducing a kind of brand-new, based on Computer Vision book contents recognizes match party
Method.
The purpose that the present invention is adopted is realized at least through one of following technical scheme.
A kind of book contents searching and matching method based on Computer Vision, by combining based on the target of image procossing
Image segmentation algorithm and target image is split and is extracted based on the target image extraction algorithm of rim detection, and passed through
Based on the target image search matching algorithm for perceiving hash algorithm, matching is scanned for target image, specifically include following step
Suddenly：
（1）, book image partitioning algorithm based on image procossing, image is converted to the video sequence that camera is captured, and
Book image region is set, then input picture is split with image segmentation algorithm, obtain target image region；
（2）, book image extraction algorithm based on four rim detections, the image segmentation for splitting is converted to into gray-scale map, and
Target image edge is extracted based on four edge detection algorithms, and deducts pixel boundary, extract target image；
（3）, based on perceive hash algorithm book contents picture search matching algorithm, by the way that whole database object image is entered
Row perceives Hash and encodes and preserve coding, then obtains the Hash coding of target image, is encoded and data by calculating target image
Hamming distance between the Image Coding of storehouse come realize book contents search matching.
Further, step（1）In the target image partitioning algorithm based on image procossing, camera is captured
Video Quality Metric is the image of consolidation form, and it is that image starting and termination coordinate deduct p pixel and obtain square to arrange target area
Shape region S is foreground target region, and here p takes empirical value 10, and the back of the body is then set as within input picture outside rectangular area
Scene area, can split with reference to image segmentation algorithm according to prospect, background area and obtain foreground target region, then based on image at
The Target Segmentation algorithm steps of reason are as follows：
1）, be converted to the video sequence that camera is captured the image of consolidation form；
2）, sets target zone boundary number of pixels p be 10, and according to capture book image starting, terminate coordinate deduct p picture
Element obtains foreground target place rectangular area S, beyond the S of rectangular area, background area is then judged within capture images；
3）, prospect mesh obtained with reference to image segmentation algorithm segmentation according to capture images foreground area and background area
Mark region.
Further, step（2）It is described to be based in the book contents image zooming-out algorithm of four rim detections, work as target image
When extracting from input picture, the background black circle portion of image is removed using the target image extraction algorithm of four rim detections
Point, gray-scale map is first converted the image into, then it is background black circle part if now image pixel value 0, it is not prospect mesh for 0
Mark part, according to the method for discrimination most going up and most right two points A, B for gray-scale map can be detected, further according to coordinate formula is corrected
Rotational correction is carried out to image；Then from the middle of the four edges of image it is whether toward image inside detection image pixel value respectively again
0, be 0 and be judged to background black circle part, it is non-zero, be judged to object-image section, detection terminate to obtain the row of target image,
Row origin coordinates startRow, startCol and termination coordinate endRow, endCol, then origin coordinates and end coordinate are respectively subtracted
Boundary value pad is removed, the region of book image is obtained, pad here takes empirical value 2, then the target image based on four rim detections
Extraction algorithm step is as follows：
1）, convert the image into gray-scale map；
2）, in detection image book contents region most go up and most right two point A and B coordinate, according to the coordinate of A and B and
Correction formula carries out rotational correction to image；
3）, respectively from image up and down in the middle of four edges toward internal detection pixel value, if pixel value is not 0, obtain mesh
The origin coordinates startRow of logo image, startCol and termination coordinate endRow, endCol, otherwise continue executing with step 2）；
4）, by target image origin coordinates startRow, startCol and terminate coordinate endRow, endCol is individually subtracted border
Value pad obtains the coordinate of book image, obtains horizontal book image.
Further, step（3）It is described to be searched in matching algorithm, by database based on the target image for perceiving hash algorithm
In image respectively through step（1）And step（2）Extraction obtains target image, is first converted to gray-scale map, be compressed to
12x12 sizes simultaneously carry out discrete cosine transform and calculate discrete cosine coefficient, then carry out Hash to the region of upper left corner 8x8 and encode
To picture fingerprint and it is stored in fingerprint matrices, when matching picture to be searched is input into, input picture is carried out into again above-mentioned steps
Input picture Hash fingerprint is obtained, each fingerprint in Hash fingerprint and fingerprint matrices is carried out into Hamming distance calculating, Hamming
The image that image corresponding to the minimum fingerprint of distance is then obtained for search matching, then based on the target image for perceiving hash algorithm
Search matching algorithm step is as follows：
1）, each image in database is extracted by the target image of target image partitioning algorithm and four rim detections and is calculated
Method obtains target image；
2）, target image is converted to gray-scale map, and be compressed to the image of 12x12 and calculate discrete cosine coefficient again and obtain discrete remaining
String matrix；
3）, step 2）In discrete cosine matrix upper left corner 8x8 regions carry out Hash coding and obtain picture fingerprint and be saved in finger
In line matrix；
4）, will be input into it is to be searched matching image through step 1）, 2）, 3）64 fingerprints of target image are obtained, and it is being referred to
Hamming distance calculating is carried out in line matrix, under current search matching image is updated if Hamming distance is less than current minimum of a value
Mark, continues executing with step 4）Until search terminates.
Compared with prior art, the present invention have it is following a little and technique effect：
The present invention uses image processing techniques, and the image captured to camera carries out target image segmentation, obtains target image
Region；Again target image is extracted from object region with four edge detection algorithms, and by based on perception hash algorithm pair
Target image is encoded, and matching is scanned in database according to the coding of target image, so as to obtain in current books
The page number of appearance.The invention provides a kind of intelligentized book contents searching and matching method, is particularly suitable for children robot etc.
Education product, with extensive market prospects and practical significance.
Description of the drawings
Fig. 1 is the General Implementing method flow diagram in example.
Fig. 2 is book image partitioning algorithm flow chart；
Fig. 3 is book image extraction algorithm flow chart.
Specific embodiment
Below in conjunction with accompanying drawing and example to the present invention be embodied as be described further, but the enforcement and protection of the present invention
Not limited to this.
The technical scheme of this example mainly includes：Based on the book image partitioning algorithm of image procossing, examined based on four edges
The target image extraction algorithm of survey, it is specific as follows based on the target image search matching algorithm for perceiving hash algorithm.
1st, the book image partitioning algorithm based on Computer Vision
GrapCut algorithms based on image segmentation can carry out RGB triple channel Gaussian modelings to image,
And Target Segmentation is carried out to image by constantly carrying out the iterative process of partitioning estimation and Gauss model parameter learning.First
As object pixel M in given target given area M in the picture, region MU, as background pixel M outside the M of regionB, then distinguish
Target and background is modeled using the full covariance mixed Gauss model of a k Gaussian component, here k takes empirical value
5, in the model for obtaining each pixel can only be classified as target mixed Gauss model certain Gaussian component or background mix
Certain component of Gauss model is closed, then by optimizing mixed Gaussian to the maximum mixed Gaussian component of each pixel allocation probability
Two processes of model parameter and partitioning estimation are iterated to split target image.
Based on above-mentioned image segmentation algorithm, the present invention proposes a kind of Target Segmentation algorithm based on image procossing, leads to
Cross camera to shoot books, and the frame of video for capturing is converted to the image of the jpg forms of size 200x200.
Because the image for photographing is in addition to target image textbook, also other background images, it is necessary to it is split, sets target
Zone boundary number of pixels p is 10, and deducts p and obtain foreground target place rectangle region according to capture images starting, termination coordinate
Domain S, beyond the S of rectangular area, is then judged to background area within input picture.With the back of the body of the image segmentation algorithm to capture images
Scene area and target area are split, and obtain target area image, then the book image partitioning algorithm based on image procossing is walked
It is rapid as follows：
1）, be converted to the video sequence that camera is captured 200x200 consolidation form sizes；
2）, sets target zone boundary p be 10, and according to capture images starting terminate coordinate deduct border p
Foreground target place rectangular area S is obtained, beyond the S of rectangular area, background area is then judged within input picture；
3）, book image obtained with reference to the segmentation of figure partitioning algorithm according to input picture foreground area and background area.
2nd, the book image extraction algorithm based on four rim detections
After splitting from image to the book contents region that camera is captured, background parts pixel value is 0 in image,
Black is presented；Simultaneously as when camera shoots textbook content, textbook put may not level, cause to shoot the class for coming
This includes the anglec of rotation, and in order to extract books region image from the target image that obtains of segmentation, the present invention is carried
Go out a kind of book contents image zooming-out algorithm for being based on four rim detections to extract book contents image.Work as target image
When extracting from input picture, target book image may include the anglec of rotation, if comprising the anglec of rotation, book image
Four angular coordinates be respectively most the going up of the non-zero part of pixel value in target image, most left, most lower and most right four points, first will
Image is converted to gray-scale map to be found again on wherein off-diagonal carries out image alignment and rectifys by any two point by slope adjustment
Just.Find in this example and most go up and most right two points, be designated as A and B, coordinate is respectively (xa,ya) and (xb,yb), correct formula
It is as follows：
Image is corrected by correcting formula, then image is removed using the target image extraction algorithm of four rim detections
Background black circle part, respectively whether past image inside detection image pixel value is 0 from the middle of the four edges of image, is 0 judgement
It is non-zero for background black circle part, it is judged to object-image section, detection terminates the row, column origin coordinates for obtaining target image
StartRow, startCol and termination coordinate endRow, endCol, then origin coordinates and end coordinate are respectively deducted into boundary value
Pad, obtains the region of book image, and pad here takes empirical value 2, so as to obtain the coordinate of book contents image, can extract
Go out books content images, then the target image extraction algorithm step based on four rim detections is as follows：
1）, convert the image into gray-scale map；
2）, in detection image book contents region most go up and most right two point A and B coordinate, according to the coordinate of A and B and
Correction formula carries out rotational correction to image；
3）, respectively from image up and down in the middle of four edges toward internal detection pixel value, if pixel value is not 0, obtain mesh
The origin coordinates startRow of logo image, startCol and termination coordinate endRow, endCol, otherwise continue executing with step 2）；
4）, by target image origin coordinates startRow, startCol and terminate coordinate endRow, endCol is individually subtracted border
Value pad obtains the coordinate of book image, obtains horizontal book image.
3rd, based on the book contents search matching algorithm for perceiving hash algorithm
After book contents image is extracted from the camera image that obtains of capture, in order to identify that current book contents are
Textbook is which page, and the present invention proposes a kind of based on the book contents search matching algorithm for perceiving hash algorithm.First will extract
To book contents image be compressed to 12x12 sizes and carry out discrete cosine transform calculate discrete cosine coefficient, image pressure
The size of contracting can be according to actually being chosen, but it should more than or equal to 8x8 and unsuitable excessive.Then again to upper left corner 8x8
Region carry out Hash coding, the mean value mean of the image pixel value of the 12x12 for having compressed first is calculated, to upper left corner 8x8's
Each pixel value, if greater than mean value mean, is then set to character 1, if less than mean value mean, is then set to character 0,
So as to the Hash for obtaining 64 is encoded.In order to the search for realizing image is matched, every one page image of textbook is first carried out 64 by needs
Position Hash is encoded and protected in database, input picture is carried out into Hash coding during search matching and obtains Hash fingerprint, then will be breathed out
Uncommon fingerprint calculates Hamming distance, the minimum corresponding image of Hash fingerprint of Hamming distance with each the Hash fingerprint in database
It is judged to the image most matched with current input image, so as to obtain which page that current input image is textbook content, then base
It is as follows in the book contents search matching algorithm step for perceiving hash algorithm：
1）, each image in database is extracted by the target image of target image partitioning algorithm and four rim detections and is calculated
Method obtains target image；
2）, target image is converted to gray-scale map, and be compressed to the image of 12x12 and calculate discrete cosine coefficient again and obtain discrete remaining
String matrix；
3）, step 2）In discrete cosine matrix upper left corner 8x8 regions carry out Hash coding and obtain picture fingerprint and be saved in finger
In line matrix；
4）, will be input into it is to be searched matching image through step 1）, 2）, 3）64 fingerprints of target image are obtained, and it is being referred to
Hamming distance calculating is carried out in line matrix, under current search matching image is updated if Hamming distance is less than current minimum of a value
Mark, continues executing with step 4）.
The present invention only need to carry out carrying out textbook region front shooting using camera, using based on OpenCv exploitations
Software image/video process is carried out to the video flowing that camera is obtained.Image segmentation, rotational correction, class are carried out to shooting image
This contents extraction, perception Hash coding etc. are processed, then the Hash fingerprint that coding is obtained is searched in Hash fingerprint database
Rope is matched, and the image corresponding to the fingerprint minimum with the Hash fingerprint Hamming distance that coding is obtained then matches the figure for obtaining for search
Picture, so as to obtain the number of pages of current textbook content.
As shown in Fig. 1 General Implementing method flows, this example uses image processing techniques, the image captured to camera
Target image segmentation is carried out, object region is obtained；Again target is extracted from object region with four edge detection algorithms
Image, and by being encoded to target image based on perception hash algorithm, entered in database according to the coding of target image
Line search is matched, so as to obtain the page number of current book contents.This example by installing camera on educational robot, little
Camera shoots to book contents when child does one's assignment, and when child encounters problems, opens Video Image Segmentation and searches
Rope matching feature, the image to photographing is processed, and the content currently learnt so as to obtain child is corresponding to books
The page number, child only needed in addition say " which topic " exercise question number is obtained in conjunction with speech recognition, so as to obtain child
The sub exercise question for currently doing simultaneously can furnish an answer from backstage.
Comprise the following steps that：
First, the book image partitioning algorithm based on image procossing
First books are shot by camera, and the frame of video for capturing is converted to into the jpg lattice of size 200x200
The image of formula.Sets target zone boundary p is 10, obtains figure background area and foreground area, then is calculated with image segmentation
Method is split to the background area and target area of capture images, obtains target area image, as shown in Figure 2.
2nd, the book image extraction algorithm based on four rim detections
When target image is extracted from input picture, first convert the image into gray-scale map and find again on wherein off-diagonal
Any two point, by slope adjustment image alignment correction is carried out.Again using the target image extraction algorithm of four rim detections come
Remove the background black circle part of image, detection terminates to obtain the row, column origin coordinates startRow of target image, startCol and
Terminate coordinate endRow, endCol, then origin coordinates and end coordinate are respectively deducted into boundary value pad, obtain the area of book image
Domain, so as to extract book contents image, as shown in Figure 3.
3rd, based on the book contents search matching algorithm for perceiving hash algorithm
After extracting in the image that book contents image is obtained from camera capture, the book contents figure for first obtaining extraction
As being compressed to 12x12 sizes and carry out discrete cosine transform and calculate discrete cosine coefficient, then the region of upper left corner 8x8 is entered
Row Hash is encoded, and in order to the search for realizing image is matched, needs first every one page image of textbook to be carried out into 64 Hash codings simultaneously
In protecting database, input picture is carried out into Hash coding during search matching and obtains Hash fingerprint, then by Hash fingerprint and data
Each Hash fingerprint in storehouse calculates Hamming distance, and the minimum Hash fingerprint of Hamming distance is judged to current input image most
The image of matching, so as to which page that current input image is textbook content obtained.
Claims (4)
1. a kind of book contents searching and matching method based on Computer Vision, by combination the target figure of image procossing is based on
Target image is split and is extracted as partitioning algorithm and based on the target image extraction algorithm of rim detection, and by base
In perceive hash algorithm target image search matching algorithm, to target image scan for matching, it is characterised in that include as
Lower step：
（1）, book image partitioning algorithm based on image procossing, image is converted to the video sequence that camera is captured, and
Book image region is set, then input picture is split with image segmentation algorithm, obtain target image region；
（2）, book image extraction algorithm based on four rim detections, the image segmentation for splitting is converted to into gray-scale map, and
Target image edge is extracted based on four edge detection algorithms, and deducts pixel boundary, extract target image；
（3）, based on perceive hash algorithm book contents picture search matching algorithm, by the way that whole database object image is entered
Row perceives Hash and encodes and preserve coding, then obtains the Hash coding of target image, is encoded and data by calculating target image
Hamming distance between the Image Coding of storehouse come realize book contents search matching.
2. the book contents searching and matching method based on Computer Vision according to claim 1, it is characterised in that step
Suddenly（1）In the target image partitioning algorithm based on image procossing, the Video Quality Metric captured to camera is consolidation form
Image, and target area is set is image starting and terminates coordinate and deduct p pixel and obtain rectangular area S for foreground target institute
In region, here p takes empirical value 10, is then set as background area within input picture outside rectangular area, according to prospect, background
Region can be split with reference to image segmentation algorithm and obtain foreground target region, then the Target Segmentation algorithm steps based on image procossing
It is as follows：
1）, be converted to the video sequence that camera is captured the image of consolidation form；
2）, sets target zone boundary number of pixels p be 10, and according to capture book image starting, terminate coordinate deduct p picture
Element obtains foreground target place rectangular area S, beyond the S of rectangular area, background area is then judged within capture images；
3）, prospect mesh obtained with reference to image segmentation algorithm segmentation according to capture images foreground area and background area
Mark region.
3. the book contents searching and matching method based on Computer Vision according to claim 1, it is characterised in that step
Suddenly（2）It is described to be based in the book contents image zooming-out algorithm of four rim detections, when target image is extracted from input picture
When coming, the background black circle part of image is removed using the target image extraction algorithm of four rim detections, first converted the image into
Gray-scale map, is then background black circle part if now image pixel value 0, is not foreground target part for 0, according to the differentiation
Method can detect most going up and most right two points A, B for gray-scale map, rotation is carried out to image further according to coordinate correction formula and is rectified
Just；Then from the middle of the four edges of image toward image inside detection image pixel value whether it is respectively again 0, is 0 and is judged to the back of the body
Scape black circle part, it is non-zero, it is judged to object-image section, detection terminates the row, column origin coordinates for obtaining target image
StartRow, startCol and termination coordinate endRow, endCol, then origin coordinates and end coordinate are respectively deducted into boundary value
Pad, obtains the region of book image, and pad here takes empirical value 2, then the target image extraction algorithm based on four rim detections
Step is as follows：
1）, convert the image into gray-scale map；
2）, in detection image book contents region most go up and most right two point A and B coordinate, according to the coordinate of A and B and
Correction formula carries out rotational correction to image；
3）, respectively from image up and down in the middle of four edges toward internal detection pixel value, if pixel value is not 0, obtain mesh
The origin coordinates startRow of logo image, startCol and termination coordinate endRow, endCol, otherwise continue executing with step 2）；
4）, by target image origin coordinates startRow, startCol and terminate coordinate endRow, endCol is individually subtracted border
Value pad obtains the coordinate of book image, obtains horizontal book image.
4. the book contents searching and matching method based on Computer Vision according to claim 1, it is characterised in that step
Suddenly（3）It is described to be searched in matching algorithm, by the image in database respectively through step based on the target image for perceiving hash algorithm
Suddenly（1）And step（2）Extraction obtains target image, is first converted to gray-scale map, is compressed to 12x12 sizes and carries out and be discrete remaining
String transformation calculations discrete cosine coefficient, then Hash coding is carried out to the region of upper left corner 8x8 obtain picture fingerprint and be stored to fingerprint
In matrix, when matching picture to be searched is input into, input picture is carried out into again above-mentioned steps and obtains input picture Hash fingerprint,
Each fingerprint in Hash fingerprint and fingerprint matrices is carried out into Hamming distance calculating, corresponding to the minimum fingerprint of Hamming distance
The image that image is then obtained for search matching, then it is as follows based on the target image search matching algorithm step for perceiving hash algorithm：
1）, each image in database is extracted by the target image of target image partitioning algorithm and four rim detections and is calculated
Method obtains target image；
2）, target image is converted to gray-scale map, and be compressed to the image of 12x12 and calculate discrete cosine coefficient again and obtain discrete remaining
String matrix；
3）, step 2）In discrete cosine matrix upper left corner 8x8 regions carry out Hash coding and obtain picture fingerprint and be saved in finger
In line matrix；
4）, will be input into it is to be searched matching image through step 1）, 2）, 3）64 fingerprints of target image are obtained, and it is being referred to
Hamming distance calculating is carried out in line matrix, under current search matching image is updated if Hamming distance is less than current minimum of a value
Mark, continues executing with step 4）Until search terminates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610946349.3A CN106599028B (en) | 2016-11-02 | 2016-11-02 | Book content searching and matching method based on video image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610946349.3A CN106599028B (en) | 2016-11-02 | 2016-11-02 | Book content searching and matching method based on video image processing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106599028A true CN106599028A (en) | 2017-04-26 |
| CN106599028B CN106599028B (en) | 2020-04-28 |
Family
ID=58589683
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610946349.3A Active CN106599028B (en) | 2016-11-02 | 2016-11-02 | Book content searching and matching method based on video image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106599028B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107807998A (en) * | 2017-11-09 | 2018-03-16 | 广东小天才科技有限公司 | One kind, which is taken pictures, searches topic method, apparatus and smart machine |
| CN107870999A (en) * | 2017-11-07 | 2018-04-03 | 广东欧珀移动通信有限公司 | Multi-medium play method, device, storage medium and electronic equipment |
| CN108766077A (en) * | 2018-05-17 | 2018-11-06 | 广东小天才科技有限公司 | A kind of desk lamp, assisted learning method and device based on desk lamp |
| CN108843994A (en) * | 2018-07-10 | 2018-11-20 | 广东小天才科技有限公司 | A kind of learning interaction method and intelligent desk lamp based on intelligent desk lamp |
| CN109243215A (en) * | 2018-08-31 | 2019-01-18 | 广东小天才科技有限公司 | A kind of exchange method based on intelligent apparatus, intelligent apparatus and system |
| CN109272997A (en) * | 2018-08-31 | 2019-01-25 | 广东小天才科技有限公司 | A kind of smart machine and its projection interactive method |
| CN109344276A (en) * | 2018-08-27 | 2019-02-15 | 广州企图腾科技有限公司 | A kind of finger image generation method, image similarity comparative approach and storage medium |
| CN110173752A (en) * | 2018-06-19 | 2019-08-27 | 葛高丽 | Intelligent humidification-type heater |
| CN110427895A (en) * | 2019-08-06 | 2019-11-08 | 李震 | A kind of video content similarity method of discrimination based on computer vision and system |
| CN110765997A (en) * | 2019-11-11 | 2020-02-07 | 深圳一块互动网络技术有限公司 | Interactive reading implementation method |
| CN111382298A (en) * | 2018-12-30 | 2020-07-07 | 贝壳技术有限公司 | Image retrieval method and device based on picture content and electronic equipment |
| CN112115282A (en) * | 2020-09-17 | 2020-12-22 | 北京达佳互联信息技术有限公司 | Question answering method, device, equipment and storage medium based on search |
| CN113704532A (en) * | 2020-11-25 | 2021-11-26 | 天翼智慧家庭科技有限公司 | Method and system for improving recall rate of picture retrieval |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003289435A (en) * | 2002-01-25 | 2003-10-10 | Osaka Industrial Promotion Organization | Image data generation method, falsification detecting method, image data generation apparatus, falsification detection apparatus, computer program, and recording medium |
| CN101364267A (en) * | 2007-08-09 | 2009-02-11 | 富士通株式会社 | Character recognition apparatus, character recognition method, and computer product |
| CN103929644A (en) * | 2014-04-01 | 2014-07-16 | Tcl集团股份有限公司 | Video fingerprint database building method and device and video fingerprint recognition method and device |
| US20140298012A1 (en) * | 2010-09-20 | 2014-10-02 | Security First Corp. | Systems and methods for secure data sharing |
| CN105678301A (en) * | 2016-01-05 | 2016-06-15 | 珠海全志科技股份有限公司 | Method, system and device for automatically identifying and segmenting text image |
| CN105844669A (en) * | 2016-03-28 | 2016-08-10 | 华中科技大学 | Video target real-time tracking method based on partial Hash features |
| CN105912643A (en) * | 2016-04-08 | 2016-08-31 | 浙江理工大学 | Image retrieval method based on content improved Average Hash |
-
2016
- 2016-11-02 CN CN201610946349.3A patent/CN106599028B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003289435A (en) * | 2002-01-25 | 2003-10-10 | Osaka Industrial Promotion Organization | Image data generation method, falsification detecting method, image data generation apparatus, falsification detection apparatus, computer program, and recording medium |
| CN101364267A (en) * | 2007-08-09 | 2009-02-11 | 富士通株式会社 | Character recognition apparatus, character recognition method, and computer product |
| US20140298012A1 (en) * | 2010-09-20 | 2014-10-02 | Security First Corp. | Systems and methods for secure data sharing |
| CN103929644A (en) * | 2014-04-01 | 2014-07-16 | Tcl集团股份有限公司 | Video fingerprint database building method and device and video fingerprint recognition method and device |
| CN105678301A (en) * | 2016-01-05 | 2016-06-15 | 珠海全志科技股份有限公司 | Method, system and device for automatically identifying and segmenting text image |
| CN105844669A (en) * | 2016-03-28 | 2016-08-10 | 华中科技大学 | Video target real-time tracking method based on partial Hash features |
| CN105912643A (en) * | 2016-04-08 | 2016-08-31 | 浙江理工大学 | Image retrieval method based on content improved Average Hash |
Non-Patent Citations (1)
| Title |
|---|
| 李鹏远: "《图像检索算法研究及其在互联网教育中的应用》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107870999B (en) * | 2017-11-07 | 2020-09-08 | Oppo广东移动通信有限公司 | Multimedia playing method, device, storage medium and electronic equipment |
| CN107870999A (en) * | 2017-11-07 | 2018-04-03 | 广东欧珀移动通信有限公司 | Multi-medium play method, device, storage medium and electronic equipment |
| CN107807998A (en) * | 2017-11-09 | 2018-03-16 | 广东小天才科技有限公司 | One kind, which is taken pictures, searches topic method, apparatus and smart machine |
| CN108766077A (en) * | 2018-05-17 | 2018-11-06 | 广东小天才科技有限公司 | A kind of desk lamp, assisted learning method and device based on desk lamp |
| CN110173752B (en) * | 2018-06-19 | 2021-04-13 | 安徽新大陆特种涂料有限责任公司 | Intelligent humidifying type warmer |
| CN110173752A (en) * | 2018-06-19 | 2019-08-27 | 葛高丽 | Intelligent humidification-type heater |
| WO2019242386A1 (en) * | 2018-06-19 | 2019-12-26 | 葛高丽 | Smart humidification-type heater |
| CN108843994A (en) * | 2018-07-10 | 2018-11-20 | 广东小天才科技有限公司 | A kind of learning interaction method and intelligent desk lamp based on intelligent desk lamp |
| CN109344276A (en) * | 2018-08-27 | 2019-02-15 | 广州企图腾科技有限公司 | A kind of finger image generation method, image similarity comparative approach and storage medium |
| CN109344276B (en) * | 2018-08-27 | 2021-01-05 | 广州企图腾科技有限公司 | Image fingerprint generation method, image similarity comparison method and storage medium |
| CN109272997A (en) * | 2018-08-31 | 2019-01-25 | 广东小天才科技有限公司 | A kind of smart machine and its projection interactive method |
| CN109243215A (en) * | 2018-08-31 | 2019-01-18 | 广东小天才科技有限公司 | A kind of exchange method based on intelligent apparatus, intelligent apparatus and system |
| CN111382298A (en) * | 2018-12-30 | 2020-07-07 | 贝壳技术有限公司 | Image retrieval method and device based on picture content and electronic equipment |
| CN111382298B (en) * | 2018-12-30 | 2021-04-20 | 北京房江湖科技有限公司 | Image retrieval method and device based on picture content and electronic equipment |
| CN110427895A (en) * | 2019-08-06 | 2019-11-08 | 李震 | A kind of video content similarity method of discrimination based on computer vision and system |
| CN110765997A (en) * | 2019-11-11 | 2020-02-07 | 深圳一块互动网络技术有限公司 | Interactive reading implementation method |
| CN110765997B (en) * | 2019-11-11 | 2023-12-26 | 深圳一块互动网络技术有限公司 | Interactive reading realization method |
| CN112115282A (en) * | 2020-09-17 | 2020-12-22 | 北京达佳互联信息技术有限公司 | Question answering method, device, equipment and storage medium based on search |
| CN113704532A (en) * | 2020-11-25 | 2021-11-26 | 天翼智慧家庭科技有限公司 | Method and system for improving recall rate of picture retrieval |
| CN113704532B (en) * | 2020-11-25 | 2024-04-26 | 天翼数字生活科技有限公司 | Method and system for improving picture retrieval recall rate |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106599028B (en) | 2020-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106599028A (en) | Book content searching and matching method based on video image processing | |
| CN112287866B (en) | Human body action recognition method and device based on human body key points | |
| CN110738101B (en) | Behavior recognition method, behavior recognition device and computer-readable storage medium | |
| EP3916627A1 (en) | Living body detection method based on facial recognition, and electronic device and storage medium | |
| CN103530599B (en) | The detection method and system of a kind of real human face and picture face | |
| US20160154469A1 (en) | Mid-air gesture input method and apparatus | |
| CN103927016B (en) | Real-time three-dimensional double-hand gesture recognition method and system based on binocular vision | |
| CN110443205B (en) | Hand image segmentation method and device | |
| CN103020992B (en) | A kind of video image conspicuousness detection method based on motion color-associations | |
| CN104809481A (en) | Natural scene text detection method based on adaptive color clustering | |
| CN110472625B (en) | Chinese chess piece visual identification method based on Fourier descriptor | |
| CN103440633B (en) | A kind of digital picture dispels the method for spot automatically | |
| CN102147867B (en) | Method for identifying traditional Chinese painting images and calligraphy images based on subject | |
| CN110032932B (en) | Human body posture identification method based on video processing and decision tree set threshold | |
| CN104281839A (en) | Body posture identification method and device | |
| CN108304839B (en) | Image data processing method and device | |
| CN107944451B (en) | Line segmentation method and system for ancient Tibetan book documents | |
| CN104102904A (en) | Static gesture identification method | |
| CN107844742A (en) | Facial image glasses minimizing technology, device and storage medium | |
| CN106960175A (en) | The first visual angle dynamic gesture detection method based on depth convolutional neural networks | |
| CN107122775A (en) | A kind of Android mobile phone identity card character identifying method of feature based matching | |
| CN112287867A (en) | Multi-camera human body action recognition method and device | |
| CN112906550A (en) | Static gesture recognition method based on watershed transformation | |
| Tsagaris et al. | Colour space comparison for skin detection in finger gesture recognition | |
| CN106778621A (en) | Facial expression recognizing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |