CN109034154A - The extraction and recognition methods of Invoice Seal duty paragraph - Google Patents

The extraction and recognition methods of Invoice Seal duty paragraph Download PDF

Info

Publication number
CN109034154A
CN109034154A CN201810811435.2A CN201810811435A CN109034154A CN 109034154 A CN109034154 A CN 109034154A CN 201810811435 A CN201810811435 A CN 201810811435A CN 109034154 A CN109034154 A CN 109034154A
Authority
CN
China
Prior art keywords
image
duty paragraph
value
invoice seal
invoice
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
Application number
CN201810811435.2A
Other languages
Chinese (zh)
Inventor
张文博
王凯
姬红兵
李林
臧博
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Kunshan Innovation Institute of Xidian University
Original Assignee
Kunshan Innovation Institute of Xidian University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Kunshan Innovation Institute of Xidian University filed Critical Kunshan Innovation Institute of Xidian University
Priority to CN201810811435.2A priority Critical patent/CN109034154A/en
Publication of CN109034154A publication Critical patent/CN109034154A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/63—Scene text, e.g. street names
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition
    • G06V30/14—Image acquisition
    • G06V30/148—Segmentation of character regions
    • G06V30/153—Segmentation of character regions using recognition of characters or words
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
    • G06V30/10—Character recognition

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses the extraction and recognition methods of a kind of Invoice Seal duty paragraph, mainly solve the problems, such as that existing method inefficient, precision when obtaining duty paragraph are low.Implementation step includes: 1) to obtain the RGB image comprising Invoice Seal；2) RGB image is subjected to laplacian spectral radius, and is transformed into YCrCb color space；3) adaptive thresholding is carried out to the channel Cr of YCrCb color space, only retains the content of Invoice Seal；4) the outer profile elliptical image of Invoice Seal is obtained by Canny edge detection；5) by calculating the center-of-mass coordinate of elliptical image, central point is found；6) according to central point interception only comprising the rectangular area of duty paragraph；7) duty paragraph in rectangular area is divided into individual digit；8) template matching is carried out to each number, identifies duty paragraph.The present invention effectively improves the acquisition speed of Invoice Seal duty paragraph, reduces identification error, can be used for quickly accurate extraction and identification duty paragraph.

Description

The extraction and recognition methods of Invoice Seal duty paragraph
Technical field
The invention belongs to digital image processing techniques fields, further to character recognition technology, specially a kind of invoice The extraction and recognition methods of special seal duty paragraph, can be used for obtaining tax registration information.
Background technique
Invoice Seal, which refers to, to be scribed using the enterprise of invoice, unit and individual businessman by tax authority's regulation, The special seal on invoice is covered when neck is purchased or issued invoice, and contains its company's organization, invoice in the seal stamp Special seal printed words and the tax registration number, wherein tax registration abbreviation duty paragraph.Duty paragraph is the tax " body that licence issuing authority provides Part card ", unified duty paragraph system are conducive to strengthen the tax registration, and pay taxes file administration and tax source management are reinforced in specification tax collection and administration, Promotion is managed the tax in accordance with the law.Therefore the extraction, identification of duty paragraph are particularly important.
At this stage there are mainly two types of the extractions and recognition methods of duty paragraph: first is that traditional manual identified method, passes through people Work recognizes duty paragraph, is then input in computer, and this method low efficiency, error are big, and artificial and time cost is excessively high；Two It is extraction and recognition methods based on deep learning, by a large amount of known training sample training classifiers, then using circulation mind Pretreated image digitization is positioned through network, the image with number is partitioned into, finally send the image divided To classify into trained classifier, this method can be effectively improved the problem that manual identified error is big, at high cost, however Its used Recognition with Recurrent Neural Network and classifier be required to extract image specific characteristic as input, therefore calculation amount compared with Greatly, extraction and the slow deficiency of recognition speed are still remained.
Summary of the invention
It is an object of the invention in view of the above shortcomings of the prior art, propose a kind of extraction of Invoice Seal duty paragraph and Recognition methods quickly positions Invoice Seal centre coordinate, and then selected duty paragraph location by calculating the mass center of elliptical image Then domain is partitioned into individual digit image, successively carry out template matching to it, rapidly and accurately identifies the tax in Invoice Seal Number, to improve the recognition efficiency and precision to Invoice Seal duty paragraph.
To achieve the above object, the technical scheme adopted by the invention is as follows:
(1) RGB image comprising Invoice Seal is obtained；
(2) RGB image that will acquire carries out laplacian spectral radius, then converts it to YCrCb color space；
(3) YCrCb color space is subjected to channel separation, and adaptive thresholding is carried out to the channel Cr isolated, Binary map is converted the image into, then removes the noise of binary map by dilation erosion operation, obtains only including Invoice Seal Binary map；
(4) Canny edge detection is carried out to the binary map for only including Invoice Seal, selects outermost layer profile, i.e. invoice The outer profile elliptical image of special seal；
(5) center-of-mass coordinate of outer profile elliptical image, the i.e. center point coordinate of Invoice Seal are calculated；
(6) according to center point coordinate interception only comprising the rectangular area of duty paragraph；
(7) duty paragraph in rectangular area is divided into individual digit；
(8) the search pattern T of production number 0~9, adjusts the size of individual digit image, makes itself and number in search pattern T Word it is in the same size, be sequentially sent in the numeric search template made, adopt using individual digit adjusted as searched figure S Number is matched with correlation method, duty paragraph is identified according to matching result.
Compared with prior art, the present invention having the advantage that
The first, it is sat since the present invention obtains the elliptical central point of Invoice Seal outer profile using zero and first order geometric moment Then mark is accurately positioned duty paragraph position according to center point coordinate, there is no need to training network and characteristics of image is extracted, to reduce The calculation amount of duty paragraph positioning improves setting accuracy.
The second, the numerical portion in Invoice Seal is identified using the method for template matching due to the present invention, because This does not need to train classifier and extracts the process of characteristics of image, to accelerate the recognition speed to number, improve Take and identify the efficiency of duty paragraph.
Detailed description of the invention
Fig. 1 is implementation flow chart of the invention；
Fig. 2 is that the RGB comprising Invoice Seal schemes；
Fig. 3 is the binary map for only including Invoice Seal content；
Fig. 4 is the outer profile image of Invoice Seal；
Fig. 5 is Invoice Seal outer profile and scattergram picture；
Fig. 6 is the rectangle binary map for only including duty paragraph；
Fig. 7 is to carry out the image obtained after color inversion to the rectangle binary map for only including duty paragraph.
Specific embodiment
Technical solution in order to enable those skilled in the art to better understand the present invention, with reference to the accompanying drawing with specific implementation Example, is further elaborated the present invention.
Referring to Fig.1, the extraction and recognition methods of Invoice Seal duty paragraph, comprising the following steps:
Step 1. obtains the RGB image comprising Invoice Seal, as shown in Figure 2.
The RGB image that step 2. will acquire carries out laplacian spectral radius, then converts it to YCrCb color space.
Firstly, laplacian spectral radius processing is carried out to the RGB image that step 1 obtains, it is specific as follows: to will be less than neighborhood territory pixel The center pixel gray value of average gray value reduces, and the center pixel gray value that will be above neighborhood territory pixel average gray value improves. That is, reducing the ash of this center pixel when centre of neighbourhood pixel grey scale is lower than the average gray of other pixels in field where it Angle value；When centre of neighbourhood pixel grey scale is higher than the average gray of other pixels in its place neighborhood, this center pixel is improved Gray value.After Edge contrast, Invoice Seal part is easier to extract image.
Then, the RGB image after sharpening is transformed by YCrCb color space by following formula:
Y=(B × 1868+G × 9617+R × 4899+8192) ÷ 16384；
Cr=((B-Y) × 9241+8192) ÷ 16384+128；
Cb=((R-Y) × 11682+8192) ÷ 16384+128；
Wherein, " R " indicates the value in the channel R in RGB image, and " G " indicates the value in the channel G in RGB image, and " B " indicates RGB figure The value of channel B as in, " Y " indicate the value in the channel Y in YCrCb image, and " Cr " indicates the value in the channel Cr in YCrCb image, " Cb " Indicate the value in the channel Cb in YCrCb image.
YCrCb color space includes three channels, in which: " Y " indicates brightness, i.e. grayscale value；" Cr " and " Cb " equal table Show coloration, " Cr " refers to red chrominance component, and " Cb " refers to chroma blue component.
Step 3. carries out adaptive thresholding to the channel Cr of YCrCb color space, only retains the interior of Invoice Seal Hold.
The channel Cr is isolated from YCrCb color space, and adaptive thresholding is carried out to the channel Cr, is converted the image into Then binary map carries out dilation erosion operation removal noise to it, the binary map for only retaining Invoice Seal is obtained, such as Fig. 3 institute Show.
Step 4. obtains the outer profile elliptical image of Invoice Seal by Canny edge detection.
Canny edge detection is carried out to the binary map for only retaining Invoice Seal by following steps:
1) Gaussian filter is used, smoothed image filters out noise；
2) gradient intensity of each pixel and direction in image are calculated；
3) edge detection bring spurious response is eliminated using non-maxima suppression；
4) it is detected using dual threshold and determines true and potential edge；
5) by inhibiting isolated weak edge to be finally completed edge detection.
Outermost layer profile, i.e. the outer profile elliptical image of Invoice Seal are selected after completing edge detection, as shown in Figure 4.
Step 5. finds the central point of Invoice Seal by the center-of-mass coordinate of calculating elliptical image.
According to national standard, Invoice Seal duty paragraph position in its center, it is therefore desirable to find the center of Invoice Seal Point, for positioning duty paragraph position.There are many modes for calculating the center-of-mass coordinate of outer profile elliptical image, as geometric moment, Hough become It changes, the present invention calculates the center-of-mass coordinate for obtaining outer profile elliptical image by the way of geometric moment, the specific steps are as follows:
The p+q rank geometric moment m of discrete picturepqIs defined as:
Wherein, x, y respectively indicate the transverse and longitudinal coordinate of outer profile elliptical image；P indicates that the order of x, q indicate the order of y, and P, q is the integer more than or equal to 0；M indicates the boundary of outer profile elliptical image abscissa x, and N indicates that outer profile elliptical image is vertical and sits The boundary of y is marked, and M, N are the integer more than or equal to 1；
P=0, q=0 are taken, zeroth order geometric moment m is obtained according to formula<1>00:
P=1, q=0 are taken, x single order geometric moment m is obtained according to formula<1>10:
P=0, q=1 are taken, y single order geometric moment m is obtained according to formula<1>01:
Using zero and first order geometric moment, sat by the mass center that following formula acquires Invoice Seal outer profile elliptical image Mark
The central point of Invoice Seal outer profile is as shown in Figure 5.
If calculating the center-of-mass coordinate of outer profile elliptical image using Hough transformation in the step, following steps are needed:
Conic section in plane is indicated are as follows:
Ax2+2Bxy+Cy2+ 2Dx+2Ey+1=0
Wherein, A, B, C, D, E indicate five parameters to be asked, and x, y indicate the transverse and longitudinal coordinate of conic section.
Above five parameters are found out using Hough transformation, if B2- AC < 0, then conic section is ellipse, then according to ellipse The elliptical centre coordinate of Circle Parameters equation solution.
Step 6. is according to center point coordinate interception only comprising the rectangular area of duty paragraph.
According to Invoice Seal sample chapter standard, duty paragraph, can be according to center point coordinate and adaptive in the middle position of seal Threshold value is answered, the minimum and complete rectangular area comprising duty paragraph is taken out, i.e., only includes the rectangle binary map of duty paragraph, as shown in Figure 6.
Duty paragraph is divided into individual digit by step 7..
Rectangular region image is subjected to color inversion by what step 6 extracted, such as Fig. 7 of the image after obtaining color inversion It is shown, further according to adaptively threshold value is determined, image after reversion is split, segmentation obtains each number.It will be in rectangular area Duty paragraph is divided into individual digit.
Step 8. carries out template matching to each number, identifies duty paragraph.
The search pattern T of production number 0~9, adjusts the size of individual digit image, makes itself and number in search pattern T It is in the same size, be sequentially sent in the numeric search template made, use using individual digit adjusted as searched figure S Correlation method matches number, identifies duty paragraph according to matching result.
Carry out that matched detailed process is as follows to number using correlation method:
(8.1) setting search pattern T has m × n pixel, searched figure S to have W × H pixel, and T is overlayed on S simultaneously It is translated, the region S under T covering is subgraph Sij, search pattern T and subgraph S are obtained according to the following formulaijSimilarity D (i, j):
Wherein, i is abscissa of the subgraph upper left corner on searched figure S, and j is the subgraph upper left corner on searched figure S Ordinate, and 1≤i≤W-m, 1≤j≤H-n, Sij(m, n) is SijPixel value at coordinate (m, n), T (m, n) are that T is being sat Mark the pixel value at (m, n)；
(8.2) D (i, j) is normalized, obtains the coefficient R (i, j) of template matching:
(8.3) coefficient R (i, j)=1 indicates that template and subgraph are completely the same, due to that can have certain deviation, we It provides herein, if R (i, j), which is more than or equal to 0.9, indicates that template is consistent with subgraph.Judge whether the value of coefficient R (i, j) is big In being equal to 0.9, if so, subgraph is consistent with template, matching search is completed, determines the corresponding number of picture；Conversely, subgraph and mould Plate is inconsistent, continues iterative search, until searching and the consistent template of subgraph, completion matching search.
Unspecified part of the present invention belongs to common sense well known to those skilled in the art.
Above description is only the embodiment of the present invention, does not constitute any limitation of the invention, it is clear that for this field For professional, after having understood the content of present invention and principle, it may all carry out without departing from the principles of the invention Various modifications and variations in form and details, but these modifications and variations based on inventive concept are still in power of the invention Within the protection scope that benefit requires.

Claims (9)

1. a kind of extraction and recognition methods of Invoice Seal duty paragraph, which comprises the steps of:
(1) RGB image comprising Invoice Seal is obtained；
(2) RGB image that will acquire carries out laplacian spectral radius, then converts it to YCrCb color space；
(3) YCrCb color space is subjected to channel separation, and adaptive thresholding is carried out to the channel Cr isolated, will schemed Picture is converted to binary map, then the noise of binary map is removed by dilation erosion operation, obtains two-value only comprising Invoice Seal Figure；
(4) Canny edge detection is carried out to the binary map for only including Invoice Seal, selects outermost layer profile, i.e. invoice is dedicated The outer profile elliptical image of chapter；
(5) center-of-mass coordinate of outer profile elliptical image, the i.e. center point coordinate of Invoice Seal are calculated；
(6) according to center point coordinate interception only comprising the rectangular area of duty paragraph；
(7) duty paragraph in rectangular area is divided into individual digit；
(8) production number 0~9 search pattern T, adjust individual digit image size, make its with it is digital in search pattern T It is in the same size, it is sequentially sent to individual digit adjusted as searched figure S in the numeric search template made, using phase Pass method matches number, identifies duty paragraph according to matching result.
2. according to the method described in claim 1, it is characterized by: carrying out Laplce to the RGB image of acquisition in step (2) Edge contrast, specific processing mode are as follows: the center pixel gray value that will be less than neighborhood territory pixel average gray value reduces, and will be above neighbour The center pixel gray value of domain pixel average gray value improves.
3. according to the method described in claim 1, it is characterized by: the RGB image after sharpening is passed through following public affairs in step (2) Formula is transformed into YCrCb color space:
Y=(B × 1868+G × 9617+R × 4899+8192) ÷ 16384；
Cr=((B-Y) × 9241+8192) ÷ 16384+128；
Cb=((R-Y) × 11682+8192) ÷ 16384+128；
Wherein, " Y " indicates the value in the channel Y in YCrCb image, and " Cr " indicates the value in the channel Cr in YCrCb image, and " Cb " is indicated The value in the channel Cb in YCrCb image, " R " indicate the value in the channel R in RGB image, and " G " indicates the value in the channel G in RGB image, " B " Indicate the value of channel B in RGB image.
4. according to the method described in claim 1, it is characterized by: to the binary map for only including Invoice Seal in step (4) Canny edge detection is carried out, is realized especially by following steps:
(4.1) binary map is denoised using Gaussian filter；
(4.2) gradient intensity of each pixel and direction in image are calculated after denoising；
(4.3) edge detection bring spurious response is eliminated by the way of non-maxima suppression；
(4.4) it is detected using dual threshold and determines true and potential edge；
(4.5) by inhibiting isolated weak edge to be finally completed edge detection.
5. according to the method described in claim 1, it is characterized by: the center-of-mass coordinate of outer profile elliptical image is logical in step (5) It crosses geometric moment or Hough transformation is calculated.
6. according to the method described in claim 5, it is characterized by: being sat by the mass center that geometric moment calculates outer profile elliptical image Mark, the specific steps are as follows:
The p+q rank geometric moment m of discrete picturepqIs defined as:
Wherein, x, y respectively indicate the transverse and longitudinal coordinate of outer profile elliptical image；P indicates that the order of x, q indicate the order of y, and p, q It is the integer more than or equal to 0；M indicates the boundary of outer profile elliptical image abscissa x, and N indicates outer profile elliptical image ordinate y Boundary, and M, N are the integer more than or equal to 1；
P=0, q=0 are taken, zeroth order geometric moment m is obtained according to formula<1>00:
P=1, q=0 are taken, x single order geometric moment m is obtained according to formula<1>10:
P=0, q=1 are taken, y single order geometric moment m is obtained according to formula<1>01:
Using zero and first order geometric moment, the center-of-mass coordinate of Invoice Seal outer profile elliptical image is acquired by following formula
7. according to the method described in claim 1, it is characterized by: only the rectangular area comprising duty paragraph is basis in step (6) Center-of-mass coordinate and the minimum of adaptive threshold taking-up and the complete region comprising duty paragraph.
8. according to the method described in claim 1, it is characterized by: the duty paragraph in rectangular area is divided into list in step (7) A number, including operate as follows:
(7.1) rectangular region image is subjected to color inversion, image after being inverted；
(7.2) according to adaptive threshold, image after reversion is split, obtains individual digit image.
9. according to the method described in claim 1, it is characterized by: matched to number progress using correlation method in step (8) Detailed process is as follows:
(8.1) setting search pattern T has m × n pixel, searched figure S to have W × H pixel, and T is overlayed on S and carried out Translation, the region S under T covering are subgraph Sij, search pattern T and subgraph S are obtained according to the following formulaijSimilarity D (i, j):
Wherein, i is abscissa of the subgraph upper left corner on searched figure S, and j is vertical seat of the subgraph upper left corner on searched figure S Mark, and 1≤i≤W-m, 1≤j≤H-n, Sij(m, n) is SijPixel value at coordinate (m, n), T (m, n) be T coordinate (m, N) pixel value at place；
(8.2) D (i, j) is normalized, obtains the coefficient R (i, j) of template matching:
(8.3) judge whether the value of coefficient R (i, j) is more than or equal to 0.9, if so, subgraph is consistent with template, complete matching Search, determines the corresponding number of picture；Conversely, subgraph and template are inconsistent, continue iterative search, until searching and subgraph one The template of cause completes matching search.
CN201810811435.2A 2018-07-23 2018-07-23 The extraction and recognition methods of Invoice Seal duty paragraph Pending CN109034154A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810811435.2A CN109034154A (en) 2018-07-23 2018-07-23 The extraction and recognition methods of Invoice Seal duty paragraph

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810811435.2A CN109034154A (en) 2018-07-23 2018-07-23 The extraction and recognition methods of Invoice Seal duty paragraph

Publications (1)

Publication Number Publication Date
CN109034154A true CN109034154A (en) 2018-12-18

Family

ID=64645122

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810811435.2A Pending CN109034154A (en) 2018-07-23 2018-07-23 The extraction and recognition methods of Invoice Seal duty paragraph

Country Status (1)

Country Link
CN (1) CN109034154A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110598686A (en) * 2019-09-17 2019-12-20 携程计算机技术（上海）有限公司 Invoice identification method, system, electronic equipment and medium
CN111985204A (en) * 2020-07-29 2020-11-24 大连大学 Customs import and export commodity tax number prediction method
US11030450B2 (en) * 2018-05-31 2021-06-08 Vatbox, Ltd. System and method for determining originality of computer-generated images
CN113610098A (en) * 2021-08-19 2021-11-05 创优数字科技(广东)有限公司 Tax payment number identification method and device, storage medium and computer equipment

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101706873A (en) * 2009-11-27 2010-05-12 东软集团股份有限公司 Identification method and device of digital-class limitation marking
CN102426649A (en) * 2011-10-13 2012-04-25 石家庄开发区冀科双实科技有限公司 Simple steel seal digital automatic identification method with high accuracy rate
CN104112128A (en) * 2014-06-19 2014-10-22 中国工商银行股份有限公司 Digital image processing system applied to bill image character recognition and method
CN107610138A (en) * 2017-10-20 2018-01-19 四川长虹电器股份有限公司 A kind of bill seal regional sequence dividing method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101706873A (en) * 2009-11-27 2010-05-12 东软集团股份有限公司 Identification method and device of digital-class limitation marking
CN102426649A (en) * 2011-10-13 2012-04-25 石家庄开发区冀科双实科技有限公司 Simple steel seal digital automatic identification method with high accuracy rate
CN104112128A (en) * 2014-06-19 2014-10-22 中国工商银行股份有限公司 Digital image processing system applied to bill image character recognition and method
CN107610138A (en) * 2017-10-20 2018-01-19 四川长虹电器股份有限公司 A kind of bill seal regional sequence dividing method

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11030450B2 (en) * 2018-05-31 2021-06-08 Vatbox, Ltd. System and method for determining originality of computer-generated images
CN110598686A (en) * 2019-09-17 2019-12-20 携程计算机技术（上海）有限公司 Invoice identification method, system, electronic equipment and medium
CN111985204A (en) * 2020-07-29 2020-11-24 大连大学 Customs import and export commodity tax number prediction method
CN111985204B (en) * 2020-07-29 2023-06-02 大连大学 Method for predicting tax numbers of customs import and export commodities
CN113610098A (en) * 2021-08-19 2021-11-05 创优数字科技(广东)有限公司 Tax payment number identification method and device, storage medium and computer equipment

Similar Documents

Publication Publication Date Title
CN107578035B (en) Human body contour extraction method based on super-pixel-multi-color space
CN109034154A (en) The extraction and recognition methods of Invoice Seal duty paragraph
CN104751142B (en) A kind of natural scene Method for text detection based on stroke feature
CN107944452B (en) Character recognition method for circular seal
Saxena Niblack’s binarization method and its modifications to real-time applications: a review
CN109636784B (en) Image saliency target detection method based on maximum neighborhood and super-pixel segmentation
CN103942797B (en) Scene image text detection method and system based on histogram and super-pixels
CN103824091B (en) A kind of licence plate recognition method for intelligent transportation system
CN109145915A (en) License plate rapid distortion antidote under a kind of complex scene
Harraj et al. OCR accuracy improvement on document images through a novel pre-processing approach
CN104732231B (en) A kind of recognition methods of valuable bills
CN104408449A (en) Intelligent mobile terminal scene character processing method
CN110020692A (en) A kind of handwritten form separation and localization method based on block letter template
CN108629783A (en) Image partition method, system and medium based on the search of characteristics of image density peaks
CN114359998B (en) Identification method of face mask in wearing state
Turki et al. Text detection based on MSER and CNN features
CN109446984A (en) Traffic sign recognition method in natural scene
CN104123554A (en) SIFT image characteristic extraction method based on MMTD
CN112906550A (en) Static gesture recognition method based on watershed transformation
CN108711160A (en) A kind of Target Segmentation method based on HSI enhancement models
CN107729863B (en) Human finger vein recognition method
CN110084587A (en) A kind of service plate automatic settlement method based on edge context
CN112365485B (en) Melanoma identification method based on Circular LBP and color space conversion algorithm
CN108230409B (en) Image similarity quantitative analysis method based on multi-factor synthesis of color and content
Liu et al. Identification Method of Sunflower Leaf Disease Based on SIFT Point

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20181218