CN101615244A - Handwritten plate blank numbers automatic identifying method and recognition device - Google Patents
Handwritten plate blank numbers automatic identifying method and recognition device Download PDFInfo
- Publication number
- CN101615244A CN101615244A CN200810122709A CN200810122709A CN101615244A CN 101615244 A CN101615244 A CN 101615244A CN 200810122709 A CN200810122709 A CN 200810122709A CN 200810122709 A CN200810122709 A CN 200810122709A CN 101615244 A CN101615244 A CN 101615244A
- Authority
- CN
- China
- Prior art keywords
- image
- character
- slab
- cutting
- width ratio
- 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.)
- Pending
Links
Images
Landscapes
- Character Input (AREA)
Abstract
The present invention relates to a kind of number sign automatic identifying method and corresponding recognition device.This method comprises the character depth-width ratio of picked-up slab image, the corresponding kinds of characters data message of storage, and slab image, the image processing step of picked-up, treatment step comprises image transitions, figure image intensifying, denoising, binaryzation, character cutting, extraction characteristic quantity, character recognition.The present invention is not only according to the characteristics of on-the-spot hand-written character, adopting various effective measures to remove disturbs, make its sharpening, and caught each written handwriting to have the characteristics of specific rule character depth-width ratio, after character picture is carried out necessary processing, transfer the character depth-width ratio of adding up in advance according to result and carry out reasonable cutting, thereby guaranteed the accuracy rate of follow-up identification.Thereby solve the bottleneck problem that occurs in iron and steel enterprise's fully-automatic production, enhance productivity and management level.
Description
Technical field
The present invention relates to a kind of number sign automatic identifying method, the especially number automatic method of identification of the online handwriting body slab in field such as iron and steel manufacturing, and corresponding recognition device.
Background technology
Along with improving constantly and the quickening day by day of rhythm of production of steel-making, hot rolling production automation degree, one of bottleneck problem that occurs in full automatic production is the identification to handwritten form slab number, and the artificial cognition mode that falls behind has seriously hindered the raising of production capacity.
Through investigation, at present iron and steel enterprise all adopts manual type to the recognition method of slab number, promptly 24 hours manually on duty, by naked eyes identification production line slab number, manually the corresponding supervisory system of input is checked, so not only greatly reduce recognition efficiency, also occur identification mistake or typing mistake easily, it is inconsistent or imperfect to cause data to occur.
Current, existing number identification facility or equipment are divided into following three kinds usually: a kind of is the system that block letter is discerned, automatic identification as modal license plate, it is wide that this mode grasps image range, image boundary detects easily, and slab number is in the picturesque scene of a background complexity, do not have tangible border to discern, and utilizes the automatic identification equipment of license plate obviously unworkable; Another kind is to discern by writing the fashionable stroke order or the style of brushwork, as the writing pencil on computing machine and some mobile phones, significantly connects pen because slab number often exists, so this note identification apparatus is also infeasible to the identification of slab number; Also having a kind of is identification to handwritten form, Automated Sorting System as the mail in the mail system, this mode seem with the identification comparing class of slab number seemingly, but, find that through the test to a large amount of slab data in scene the temperature when writing owing to slab is than higher, hot-air causes the writing distortion, simultaneously, the result of disconnected word, fuzzy, desalination also might appear in handwritten plate blank numbers on hot slab.In this case, handwritten plate blank numbers extraction task need be studied some following contents: writing stroke was often both hasty and careless, and therefore inter-adhesive again, covering, and written word cognition generation inclination can't adopt existing equipment that the hand-written script cutting is discerned.Experiment showed, that above three kinds of modes are extremely low, all right obstructed to the discrimination of slab number.
Summary of the invention
The objective of the invention is to: the present situation that can't know handwritten plate blank numbers at above-mentioned prior art, a kind of handwritten plate blank numbers automatic identifying method is proposed, provide corresponding identification side device simultaneously, thereby solve the bottleneck problem that occurs in iron and steel enterprise's fully-automatic production, enhance productivity and management level.
In order to reach above purpose, handwritten plate blank numbers automatic identifying method of the present invention may further comprise the steps:
The picked-up step, picked-up slab image;
Storing step, the character depth-width ratio of storing corresponding kinds of characters data message, and the slab image to be identified of picked-up;
Image processing step grasps and handles the slab image of storing to be identified; Comprising:
Image transitions: grasp slab image to be identified, convert thereof into gray level image;
Figure image intensifying:, and, make the figure image intensifying to histogram equalization by the histogram of foundation conversion back gray level image;
Denoising: carry out The disposal of gentle filter to strengthening image, remove the background noise in the image;
Binaryzation: the image after the denoising is converted into bianry image;
Character cutting: bianry image is carried out the projection of level and vertical direction, obtain projection histogram, therefrom isolate character and background information, from memory storage, transfer corresponding character depth-width ratio according to isolated character data information and carry out character cutting with predetermined threshold value;
Extract characteristic quantity: extract image unchangeability characteristic quantity, directional statistics characteristic quantity and eigenvector in the character data after the cutting one by one;
Character recognition:, adopt predetermined comparison recognition methods to identify character according to the character feature amount of extracting.
The corresponding handwritten plate blank numbers automatic identification equipment of the present invention comprises camera, is used to absorb the slab image;
Memory storage, the character depth-width ratio that is used to store corresponding kinds of characters data message, and the slab image of picked-up;
Image processing apparatus is used for and will grasps and handle the slab image to be identified of storage; Wherein contain:
Image conversion apparatus is used to grasp slab image to be identified, converts thereof into gray level image;
Image intensifier device is used for the histogram by foundation conversion back gray level image, and to histogram equalization, makes the figure image intensifying;
The denoising device is used for carrying out The disposal of gentle filter to strengthening image, removes the background noise in the image;
The binaryzation device is used for the image after the denoising is converted into bianry image;
The character cutting device, be used for bianry image is carried out the projection of level and vertical direction, obtain projection histogram, therefrom isolate character and background information, from memory storage, transfer corresponding character depth-width ratio according to isolated character data information and carry out character cutting with predetermined threshold value;
Characteristic amount extraction device is used for extracting one by one image unchangeability characteristic quantity, directional statistics characteristic quantity and the eigenvector of character data after the cutting;
Character recognition device is used for according to the character feature amount of extracting, and adopts predetermined comparison recognition methods to identify hand-written character one by one.
The accurate cutting of image digitization is very crucial in the above process, because digit recognition needs single carrying out,, we can say that the accuracy of image digitization cutting will directly influence the accuracy rate of identification so must come the image that comprises a series of numerals cutting accurately.Because the handwritten form of slab number might produce adhesion, stroke covers and font tilts, therefore can not directly adopt general level and vertical syncopation, above process to image carry out level and vertical direction projection, isolate character information, and consider that each written handwriting has the character depth-width ratio of specific rule, therefore according to handling the isolated character information in back, transfer the character depth-width ratio of statistics storage in advance, can carry out the reasonable cutting of character, guarantee the accuracy rate of follow-up identification.
In a word, the present invention is not only according to the characteristics of on-the-spot hand-written character, adopting various effective measures to remove disturbs, make its sharpening, and caught each written handwriting to have the characteristics of specific rule character depth-width ratio, after character picture is carried out necessary processing, transfer the character depth-width ratio of adding up in advance according to result and carry out reasonable cutting, thereby guaranteed the accuracy rate of follow-up identification, solved the bottleneck problem that occurs in iron and steel enterprise's fully-automatic production, enhanced productivity and management level.
Description of drawings
The present invention is further illustrated below in conjunction with accompanying drawing.
Fig. 1 is the process flow diagram of one embodiment of the invention.
Fig. 2 is the image binaryzation cross-reference figure of Fig. 1 embodiment.
Fig. 3 is the image character cutting cross-reference figure of Fig. 1 embodiment.
Embodiment
Embodiment one
The handwritten plate blank numbers automatic identifying method of present embodiment comprises key steps 9 (referring to Fig. 1) such as image capture, storage, image transitions, figure image intensifying, denoising, binaryzation, character cutting, eigenvectorization, character recognition.Detailed process is as follows:
Image Acquisition and storage---by camera picked-up slab image; Slab image to be identified by the picked-up of the memory device stores of DVR and so on, and according to statistics to on-the-spot writer's actual conditions, store the character depth-width ratio of corresponding different writer's person's handwriting characteristics, and make it and reflect that its respective symbols data message of writing characteristics is corresponding, this statistics can constantly be carried out, thereby it is perfect that memory contents is enriched constantly by learning training.
Following process is finished by computing machine:
Image transitions---convert coloured image to gray level image.The image of camera collection generally all is a coloured image, and colouring information is unnecessary redundant information comparatively speaking, so before identification coloured image is converted into 8 gray level images.
General coloured image all adopts the RGB color, according to the coloud coding equation, a secondary coloured image is converted into gray level image, and its formula is:
Gray＝0.3R+0.59G+0.11B <1>
The purpose of figure image intensifying---figure image intensifying is outstanding relevant thematic information, improves the visual effect of image, and its detailed process is
Set up histogram, the histogram of digital picture is the probability statistics distribution plan of each gray level of image, the general picture that the reflection gradation of image distributes.Pixel grayscale before and after making variable r and s that representative image strengthens respectively, the probability density of corresponding grey level distribution is respectively P
r(r), P
s(s).In interval [0,1], then r=0 represents to deceive in the gray level coordinate as the normalization of pixel gray-scale value, and r=1 represents white.Any r value in interval [0,1] is carried out conversion by transforming function transformation function s=T (r), and T (r) satisfies two conditions: (1) monodrome monotone increasing function; (2) 0≤T (r)≤1.
Condition (1) keeps from deceiving to white order gray level, and grey scale pixel value in allowed limits after condition (2) guaranteed mapping transformation.Contravariant from s to r is changed to:
r＝T
-1(s)，0≤s≤1 <2>
Equally, regulation variable s also satisfy condition (1) and (2).Know by probability theory, if P
r(r) and the known T of transforming function transformation function s=T (r)
-1(s) be the monodrome monotone increasing function, then have:
Setting up histogram essence is the appearance that changes image by the probability density function of transforming function transformation function T (r) control image gray levels.
Histogram equalization claims histogram equalization again, comes down to image is carried out non-linear stretching, redistributes the image pixel value, and the quantity that makes pixel value in certain tonal range about equally.Histogram equalization process to image is finished by transforming function transformation function s=T (r), s, r be respectively pixel on target image and the original image (x, y).For consecutive image, transforming function transformation function is:
This formula the right be cumulative distribution function (Cumulative Distribution Function, CDF), differentiate has to r by this formula:
Substitution formula 2-4 obtains:
This explanation, at variable s after the conversion in field of definition, P
s(s) be even probability density.On figure image intensifying meaning, this dynamic range that is equivalent to pixel increases.
For discrete picture, gray level r
kProbable value be:
Wherein, the sum of pixel in the n presentation video, n
kBe the number of times that occurs this gray level in image, L represents the number of gray level, P
r(r
k) be the probability of k level gray level.Formula＜4 with consecutive image〉corresponding, discrete form is:
Contravariant is changed to:
r
k＝T
-1(s
k)，0≤s
k≤1 <9>
As seen, under discrete condition, can directly utilize formula＜8〉transforming function transformation function T (r from the image calculation of being given
k).
Can obtain the mapping relations of each gray level of all source images like this, according to these mapping relations source images each point pixel be carried out gradation conversion again, can finish histogram equalization source figure to each gray level of target image.
Denoising---be also referred to as the smothing filtering process to image, filtering and noise reduction method commonly used mainly contains: Gauss's smothing filtering denoising method; Mean filter denoising method; Medium filtering denoising method; Optimum filtering denoising method.Median filter can be removed the acnode noise, can not make the obscurity boundary of image, relatively is suitable for the denoising of digital picture, and therefore, present embodiment adopts median filtering method to carry out noise processed.This method is that the pixel in the neighborhood is sorted by gray level, selects the gray-scale value of the intermediate value replacement specified point (central point of window) of this group then, and for the odd number element, intermediate value is meant and sorts by size the middle numerical value in back; For the even number element, intermediate value is meant the mean value of middle two the element gray-scale values in ordering back.Concrete steps are: (1) is roamed template in image, and template center is overlapped with certain locations of pixels in the image; (2) read the gray-scale value of each respective pixel under the template, and these gray-scale values are formed a line from small to large, find out these values the insides one (or two) in the middle of coming, if two, calculating mean value then; (3) this intermediate value is composed pixel to the corresponding templates center.Above treatment step constantly repetitive cycling carries out, to reach the effect of outstanding destination object.
The gray level image binaryzation---with image binaryzation rear backdrop and all simplification of digital picture, for the accurate cutting of character lays the foundation.Adopt between maximum kind the difference method seek that optimal threshold, adaptive threshold are cut apart, watershed method is sought threshold value, by a large amount of fixed thresholds all can, its treatment effect is as shown in Figure 2.
Character cutting---because the identification of image digitization is single carrying out, so must come the picture that comprises a series of numerals cutting accurately, the accuracy of picture numeral cutting will directly influence the accuracy rate of identification.This step obtains projection histogram to the projection that image carries out level and vertical direction, chooses predetermined threshold value and isolates character and background information, and transfer corresponding character depth-width ratio according to isolated character data information from memory storage and carry out character cutting; Do normalized in case of necessity, so that next step feature extraction.The cutting effect as shown in Figure 3.
Characteristic Extraction---comprise the extraction etc. of unchangeability Characteristic Extraction, directional statistics features extraction and the vector characteristic of image.Its detailed process is:
1) conversion coefficient
The projective transformation Y-factor method Y is a kind of typical statistical pattern recognition method, it regards the character picture zone as the two-dimensional lattice figure, conversion coefficient by asking X-Y scheme is as the feature of classification, conversion commonly used has Karhuren-Loeve conversion, Walsh conversion, Fourier conversion, Rapid conversion etc., and in this employing is the Rapid conversion.
2) the unchangeability Characteristic Extraction of image
The unchangeability Characteristic Extraction of image is that image is carried out the elementary geometry conversion, asks each rank central moment constitutive characteristic amount again.Wherein, 0 rank square is the area (the pixel summation of character) of object in the image, 1 rank square can obtain character barycentric coordinates (iG, jG), or the like.
3) directional statistics features extraction
The directional statistics Characteristic Extraction is that the distribution characteristics of image pixel on eight directions extracted, and compares with the characteristics of image that will discern then, obtains the character of identification.
4) extraction of vector characteristic
By the direction that tracker wire is drawn, the selected node that has than deep camber extracts Vector Message, and stores with the form of set of coordinates.Carry out Target Recognition with vector form, reduced the yardstick of feature, thereby reduced calculated amount, improved the efficient and the accuracy of identification.
Handwritten plate blank numbers identification---specific implementation can adopt existing rule-based method, based on the method for statistics with based on neural network method.
Adopt rule-based method detailed process to be: the feature according to 0/1 matrix is described the described figured content of matrix.The present embodiment employing is carried out feature to one step an of behavior of matrix and is judged.
Can stipulate in advance that arabic numeral show that by the upper left corner such rule is then arranged in matrix: if in matrix first row two values are arranged is 1, and have between these two 1 greater than 10, then the represented numeral of this matrix is 4.When discerning with the method for rule, can adopt the structure of rule tree to judge, the rule of the corresponding matrix of each layer delegation of tree, such as, for the n row matrix, (1＜=k＜=n), the node in the tree is the form of " rule { satisfies the set of the numeral of rule } " to the rule that the k of the corresponding matrix of the k layer of tree is capable.
When this algorithm of operation, arrive leaf node according to rule along certain bar branch from root node (start node) beginning, algorithm finishes, the element in the output node set.In the present embodiment, matrix has 8 row, then need carry out the judgement of 8 steps at most and can draw recognition result.As seen, the bifurcated number of the complexity of algorithm and rule tree is irrelevant.And regular divide thin more, the bifurcated number is many more, and the discrimination of object is good more, and is few more to the traversal of tree degree of depth in the search procedure, and the accuracy of identification is just high more.
Certainly, can judge also that method and top narration are similar each step by the rule of every row.
Theory and practice proves, even the handwritten plate blank numbers automatic identifying method of present embodiment is in the hyperthermia radiation at recognition objective, target background is complicated and close with color of object, recognition objective does not have strict border differentiation and has a bad handwriting, the handwriting is blurred, exist under the mal-condition of write the two or more syllables of a word together and stroke covering, also can disturb owing to having adopted various effective measures to remove, make word symbol sharpening, and caught each written handwriting to have the characteristics of specific rule character depth-width ratio, after character picture is carried out necessary processing, transfer the character depth-width ratio of adding up in advance according to result and carry out reasonable cutting, thereby guarantee to make discrimination can reach more than 95%, reach and the close or better recognition effect of human eye identification, solve the bottleneck problem that occurs in iron and steel enterprise's fully-automatic production, improved production efficiency and management level.
Claims (5)
1. handwritten plate blank numbers automatic identifying method may further comprise the steps:
The picked-up step, picked-up slab image;
Storing step, the character depth-width ratio of storing corresponding kinds of characters data message, and the slab image to be identified of picked-up;
Image processing step grasps and handles the slab image of storing to be identified; Comprising:
Image transitions: grasp slab image to be identified, convert thereof into gray level image;
Figure image intensifying:, and, make the figure image intensifying to histogram equalization by the histogram of foundation conversion back gray level image;
Denoising: carry out The disposal of gentle filter to strengthening image, remove the background noise in the image;
Binaryzation: the image after the denoising is converted into bianry image;
Character cutting: bianry image is carried out the projection of level and vertical direction, obtain projection histogram, therefrom isolate character and background information, from memory storage, transfer corresponding character depth-width ratio according to isolated character data information and carry out character cutting with predetermined threshold value;
Extract characteristic quantity: extract image unchangeability characteristic quantity, directional statistics characteristic quantity and eigenvector in the character data after the cutting one by one;
Character recognition:, adopt predetermined comparison recognition methods to identify character one by one according to the character feature amount of extracting.
2. according to the described handwritten plate blank numbers automatic identifying method of claim 1, it is characterized in that: described denoising step adopts median filtering method.
3. according to the described handwritten plate blank numbers automatic identifying method of claim 2, it is characterized in that: described character recognition step adopts rule-based method.
4. a handwritten plate blank numbers automatic identification equipment comprises the camera that is used to absorb the slab image; It is characterized in that: also comprise
Memory storage, the character depth-width ratio that is used to store corresponding kinds of characters data message, and the slab image of picked-up;
Image processing apparatus is used for and will grasps and handle the slab image to be identified of storage; Wherein contain:
Image conversion apparatus is used to grasp slab image to be identified, converts thereof into gray level image;
Image intensifier device is used for the histogram by foundation conversion back gray level image, and to histogram equalization, makes the figure image intensifying;
The denoising device is used for carrying out The disposal of gentle filter to strengthening image, removes the background noise in the image;
The binaryzation device is used for the image after the denoising is converted into bianry image;
The character cutting device, be used for bianry image is carried out the projection of level and vertical direction, obtain projection histogram, therefrom isolate character and background information, from memory storage, transfer corresponding character depth-width ratio according to isolated character data information and carry out character cutting with predetermined threshold value;
Characteristic amount extraction device is used for extracting one by one image unchangeability characteristic quantity, directional statistics characteristic quantity and the eigenvector of character data after the cutting;
Character recognition device is used for according to the character feature amount of extracting, and adopts predetermined comparison recognition methods to identify hand-written character one by one.
5. according to the described handwritten plate blank numbers automatic identification equipment of claim 4, it is characterized in that: contain median filter in the described denoising device.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810122709A CN101615244A (en) | 2008-06-26 | 2008-06-26 | Handwritten plate blank numbers automatic identifying method and recognition device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810122709A CN101615244A (en) | 2008-06-26 | 2008-06-26 | Handwritten plate blank numbers automatic identifying method and recognition device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101615244A true CN101615244A (en) | 2009-12-30 |
Family
ID=41494876
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200810122709A Pending CN101615244A (en) | 2008-06-26 | 2008-06-26 | Handwritten plate blank numbers automatic identifying method and recognition device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101615244A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102136064A (en) * | 2011-03-24 | 2011-07-27 | 成都四方信息技术有限公司 | System for recognizing characters from image |
| CN102262775A (en) * | 2010-05-26 | 2011-11-30 | 深圳迈瑞生物医疗电子股份有限公司 | Image denoising method and device and image processing equipment |
| CN102663380A (en) * | 2012-03-30 | 2012-09-12 | 中南大学 | Method for identifying character in steel slab coding image |
| CN102663360A (en) * | 2012-03-30 | 2012-09-12 | 中南大学 | Method for automatic identifying steel slab coding and steel slab tracking system |
| CN103854024A (en) * | 2012-12-04 | 2014-06-11 | 百度国际科技（深圳）有限公司 | Method and device for extracting characters in image |
| CN104793068A (en) * | 2014-01-22 | 2015-07-22 | 佛山市顺德区顺达电脑厂有限公司 | Image acquisition-based automatic test method |
| CN104796582A (en) * | 2015-04-20 | 2015-07-22 | 中国矿业大学 | Video image denoising and enhancing method and device based on random ejection retinex |
| CN104794470A (en) * | 2015-05-04 | 2015-07-22 | 北京信息科技大学 | Method of digital acquisition and image processing for Dongba pictograph |
| CN107229932A (en) * | 2016-03-25 | 2017-10-03 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of image text and device |
| CN108537217A (en) * | 2018-04-04 | 2018-09-14 | 湖南科技大学 | Identification based on character code mark and localization method |
| CN109725561A (en) * | 2018-11-09 | 2019-05-07 | 余姚市雷阵雨电器有限公司 | Trunnion aperture field control platform |
| CN110059719A (en) * | 2019-03-18 | 2019-07-26 | 西北工业大学 | A kind of target identification method of the image moment based on Walsh transformation |
| CN111160147A (en) * | 2019-12-16 | 2020-05-15 | 华南理工大学 | Method for cutting and identifying calligraphy work image |
| CN105117740B (en) * | 2015-08-21 | 2021-06-15 | 北京旷视科技有限公司 | Font identification method and apparatus |
-
2008
- 2008-06-26 CN CN200810122709A patent/CN101615244A/en active Pending
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102262775B (en) * | 2010-05-26 | 2014-12-10 | 深圳迈瑞生物医疗电子股份有限公司 | Image denoising method and device and image processing equipment |
| CN102262775A (en) * | 2010-05-26 | 2011-11-30 | 深圳迈瑞生物医疗电子股份有限公司 | Image denoising method and device and image processing equipment |
| CN102136064A (en) * | 2011-03-24 | 2011-07-27 | 成都四方信息技术有限公司 | System for recognizing characters from image |
| CN102663360A (en) * | 2012-03-30 | 2012-09-12 | 中南大学 | Method for automatic identifying steel slab coding and steel slab tracking system |
| CN102663360B (en) * | 2012-03-30 | 2013-10-23 | 中南大学 | Method for automatic identifying steel slab coding and steel slab tracking system |
| CN102663380A (en) * | 2012-03-30 | 2012-09-12 | 中南大学 | Method for identifying character in steel slab coding image |
| CN103854024A (en) * | 2012-12-04 | 2014-06-11 | 百度国际科技（深圳）有限公司 | Method and device for extracting characters in image |
| CN104793068A (en) * | 2014-01-22 | 2015-07-22 | 佛山市顺德区顺达电脑厂有限公司 | Image acquisition-based automatic test method |
| CN104796582A (en) * | 2015-04-20 | 2015-07-22 | 中国矿业大学 | Video image denoising and enhancing method and device based on random ejection retinex |
| CN104796582B (en) * | 2015-04-20 | 2018-05-04 | 中国矿业大学 | Video image denoising and Enhancement Method and device based on random injection retinex |
| CN104794470A (en) * | 2015-05-04 | 2015-07-22 | 北京信息科技大学 | Method of digital acquisition and image processing for Dongba pictograph |
| CN104794470B (en) * | 2015-05-04 | 2017-11-17 | 北京信息科技大学 | A kind of digital acquisition and image processing method towards Dongba pictograph text |
| CN105117740B (en) * | 2015-08-21 | 2021-06-15 | 北京旷视科技有限公司 | Font identification method and apparatus |
| CN107229932A (en) * | 2016-03-25 | 2017-10-03 | 阿里巴巴集团控股有限公司 | A kind of recognition methods of image text and device |
| CN108537217A (en) * | 2018-04-04 | 2018-09-14 | 湖南科技大学 | Identification based on character code mark and localization method |
| CN108537217B (en) * | 2018-04-04 | 2021-06-25 | 湖南科技大学 | Character coding mark-based identification and positioning method |
| CN109725561A (en) * | 2018-11-09 | 2019-05-07 | 余姚市雷阵雨电器有限公司 | Trunnion aperture field control platform |
| CN109725561B (en) * | 2018-11-09 | 2020-08-18 | 浙江思密达智能装备有限公司 | Throat pipe opening field control platform |
| CN110059719A (en) * | 2019-03-18 | 2019-07-26 | 西北工业大学 | A kind of target identification method of the image moment based on Walsh transformation |
| CN110059719B (en) * | 2019-03-18 | 2022-08-09 | 西北工业大学 | Walsh transform-based target identification method for image moment |
| CN111160147A (en) * | 2019-12-16 | 2020-05-15 | 华南理工大学 | Method for cutting and identifying calligraphy work image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101615244A (en) | Handwritten plate blank numbers automatic identifying method and recognition device | |
| CN108960245B (en) | Tire mold character detection and recognition method, device, equipment and storage medium | |
| CN110059694A (en) | The intelligent identification Method of lteral data under power industry complex scene | |
| CN106096610B (en) | A kind of file and picture binary coding method based on support vector machines | |
| CN112686812B (en) | Bank card inclination correction detection method and device, readable storage medium and terminal | |
| CN104408449B (en) | Intelligent mobile terminal scene literal processing method | |
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN110619274A (en) | Identity verification method and device based on seal and signature and computer equipment | |
| CN105809121A (en) | Multi-characteristic synergic traffic sign detection and identification method | |
| CN104504383B (en) | A kind of method for detecting human face based on the colour of skin and Adaboost algorithm | |
| CN103093240A (en) | Calligraphy character identifying method | |
| CN109740572A (en) | A kind of human face in-vivo detection method based on partial color textural characteristics | |
| CN112950477A (en) | High-resolution saliency target detection method based on dual-path processing | |
| CN110738201B (en) | Self-adaptive multi-convolution neural network character recognition method based on fusion morphological characteristics | |
| CN105117741A (en) | Recognition method of calligraphy character style | |
| CN109740553A (en) | A kind of image, semantic segmentation data screening method and system based on identification | |
| CN110766016A (en) | Code spraying character recognition method based on probabilistic neural network | |
| CN110084327A (en) | Bill Handwritten Digit Recognition method and system based on the adaptive depth network in visual angle | |
| CN109741273A (en) | A kind of mobile phone photograph low-quality images automatically process and methods of marking | |
| CN109902751B (en) | Dial digital character recognition method integrating convolution neural network and half-word template matching | |
| CN109522865A (en) | A kind of characteristic weighing fusion face identification method based on deep neural network | |
| CN115082922A (en) | Water meter digital picture processing method and system based on deep learning | |
| CN115880566A (en) | Intelligent marking system based on visual analysis | |
| CN105447457A (en) | License plate character identification method based on adaptive characteristic | |
| CN110766001B (en) | Bank card number positioning and end-to-end identification method based on CNN and RNN |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20091230 |