CN106408533A - Card image extraction method and card image extraction system - Google Patents
Card image extraction method and card image extraction system Download PDFInfo
- Publication number
- CN106408533A CN106408533A CN201610818707.2A CN201610818707A CN106408533A CN 106408533 A CN106408533 A CN 106408533A CN 201610818707 A CN201610818707 A CN 201610818707A CN 106408533 A CN106408533 A CN 106408533A
- Authority
- CN
- China
- Prior art keywords
- image
- edge
- card
- detection method
- images
- 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
- 238000000605 extraction Methods 0.000 title claims abstract description 71
- 238000000034 method Methods 0.000 claims abstract description 96
- 238000001514 detection method Methods 0.000 claims abstract description 56
- 238000003708 edge detection Methods 0.000 claims abstract description 16
- 230000008569 process Effects 0.000 claims description 38
- 239000000284 extract Substances 0.000 claims description 17
- 230000015572 biosynthetic process Effects 0.000 claims description 8
- 238000003786 synthesis reaction Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 8
- 230000000877 morphologic effect Effects 0.000 claims description 7
- 230000000452 restraining effect Effects 0.000 claims description 7
- 230000000694 effects Effects 0.000 abstract description 3
- 230000002194 synthesizing effect Effects 0.000 abstract 2
- 230000003044 adaptive effect Effects 0.000 abstract 1
- 238000007781 pre-processing Methods 0.000 abstract 1
- 239000011159 matrix material Substances 0.000 description 12
- 238000011946 reduction process Methods 0.000 description 8
- 238000007688 edging Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000005286 illumination Methods 0.000 description 5
- 238000007323 disproportionation reaction Methods 0.000 description 4
- 239000000203 mixture Substances 0.000 description 4
- 238000007689 inspection Methods 0.000 description 3
- 230000005055 memory storage Effects 0.000 description 3
- 239000012141 concentrate Substances 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
Classifications
-
- G06T5/70—
-
- 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
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/10024—Color image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20061—Hough transform
Abstract
The invention discloses a card image extraction method and a card image extraction system. The method comprises the following steps: S1, preprocessing read to-be-extracted card images; S2, extracting corresponding edge images from the preprocessed card images based on a color image adaptive edge detection method and a phase consistency detection method, performing AND operation on the two currently extracted edge images and synthesizing coarsely-extracted images corresponding to the card images; S3, under an HSV space, detecting the edges of the coarsely-extracted images based on the phase consistency detection method to get final edge images; and S4, based on the edge images obtained in S3, synthesizing final extracted images of the card images through Hough-transform line detection. The pseudo edge of an image can be removed effectively, and the effect and precision of image extraction are improved. Moreover, complex background texture edges can be removed effectively, and the robustness to complex background images is good.
Description
Technical field
The present invention relates to technical field of image processing, particularly relate to a kind of carrying of the card image under complex background
Take method and system.
Background technology
Image object extractive technique is the important part of technical field of image processing, current image target extraction method
Mainly have:(1) the target image extracting method based on region, (2) target image extracting method based on edge, (3) region with
The target image extracting method that edge combines, (4) target image extracting method based on mathematical morphology, (5) are based on nerve
The target image extracting method of network, (6) target image extracting method based on SVMs, (7) target based on graph theory
Image extraction method.
But above-mentioned target image extracting method, be applied to for target image for quadrangle image when, such as identity card,
Bank card etc., i.e. card image, remain in following problems:
As usually having uncertainty due to card image data, generally along with information noise, this just significantly impacts
The problem of card image extraction accuracy, such as because the solution that card image extracts problem itself is often not uniquely, such as card
Multiple rectangular areas may be contained in image, therefore, it is difficult to distinguish the card target of our needs with unified method, such as by
Often affected larger by illumination and brightness disproportionation in card image extraction, if shade is contained in image section region, can be to image
The extraction of texture and feature impacts, thus can cause such card image extract mistake the problems such as.
Content of the invention
In view of the defect that prior art exists, the invention aims to providing a kind of extracting method of card image, should
Method can effectively remove image pseudo-edge, improves effect and the precision of image zooming-out, and the method can effectively remove simultaneously
Complex background texture edge is good to the robustness of complex background image.
To achieve these goals, technical scheme:
A kind of extracting method of card image is it is characterised in that comprise the steps:
S1, the card image to be extracted to reading pre-process;
S2, it is based respectively on color image-adaptive edge detection method, phase equalization detection method, from pretreated
Extract each self-corresponding edge image in card image, and the two width edge images currently being extracted are asked with computing after
Thick extraction image corresponding to synthesis card image;
S3, under HSV space, based on phase equalization detection method to described thick extraction image carry out rim detection, with
Obtain final edge image;
S4, based on edge image obtained in S3, the final extraction of card image is synthesized using Hough transformation detection of straight lines
Image.
Further, as the preferred version of the present invention,
In described S2, the two width edge images currently being extracted are asked with computing after synthesis card image corresponding to
The thick process extracting image comprises the steps:
S21, two obtained width edge images are removed pseudo-edge process, that is, respectively by two obtained breadths edge
Image switchs to gray level image, determines each self-corresponding edge threshold using histogram;And made an uproar by morphological image process removal
Sound point region, with respectively obtain with the edge image img11 corresponding to color image-adaptive edge detection method and with phase
Edge image img12 corresponding to bit integrity detection method;
S22, obtained edge image img11, edge image img12 are asked and computing, to obtain initial edge
Image;And calculate, by described initial edge image is carried out with minimum external world rectangle, the target confirming that card extracts corresponding to image
Image-region, by the original image of object region and card to be extracted is carried out dot product, obtains card image to be extracted
Thick extraction image.
Further, as the preferred version of the present invention,
Described S3 comprises the steps:
S31, by read in thick extraction image be transformed under hsv color space;
S32, rim detection is carried out to the thick H component extracting image based on phase equalization detection method, to obtain H component
Corresponding edge strength image and angular intensities image;And the edge strength being obtained image and angular intensities image are carried out
After superposition, carry out non-maximum restraining process, image skeletonization process obtains the edge image img21 under H component;
S33, rim detection is carried out to the thick S component extracting image based on phase equalization detection method, to obtain S component
Corresponding edge strength image and angular intensities image;And the edge strength being obtained image and angular intensities image are carried out
After superposition, carry out non-maximum restraining process, the edge image img22 obtaining under S component after image skeletonization process;
S34, described edge image img21, edge image img22 are asked or computing, after obtaining edge image img2
Asked and computing with the thick image that extracts, obtained final edge image img_z.
It is another object of the present invention to a kind of extraction system of card image will be provided it is characterised in that including:
Pretreatment unit, this pretreatment unit can pre-process to the card image to be extracted reading in;
First order extraction unit, this first order extraction unit can be based on color image-adaptive edge detection method, phase
Bit integrity detection method, extracts each self-corresponding edge image respectively from pretreated card image, and to current
The two width edge images being extracted asked with computing after synthesis card image corresponding to thick extraction image;
Edge image extraction unit, this edge image extraction unit can be under HSV space, based on phase equalization detection
Method carries out rim detection to described thick extraction image, obtains final edge image;
And second level extraction unit, this second level extraction unit can be based on the obtained edge of edge image extraction unit
Image, synthesizes the final extraction image of card image using Hough transformation detection of straight lines.
Compared with prior art, beneficial effects of the present invention:
1) it is directed to the Image semantic classification to be extracted reading in so that the lifting of image zooming-out efficiency of algorithm, very imperial palace need not be taken
Deposit memory space, be easy to real-time processing;
2) read pretreated image, be utilized respectively color image-adaptive edge detection method and phase equalization side
Method obtains two width image borders, and is asked and computing, thus obtaining the thick extraction of card as a result, it is possible to avoid image because of illumination
The error extraction causing with brightness disproportionation, can effectively remove image pseudo-edge so that the effect of image zooming-out is more excellent;
3) obtain two width edge images are switched to gray-scale map, automatically determine edge threshold using histogram, be then passed through
Less for image connected domain is removed (i.e. it is considered that being the region of noise spot) by morphological operation, can effectively remove background
Noise, improves the precision of result;
4) in HSV space, using phase equalization algorithm, edge inspection is carried out to the thick H component extracting result and S component
Survey, the edge image obtaining obtains final card extraction image using Hough transformation detection of straight lines and can effectively remove the complicated back of the body
Scape texture edge, the image robustness for complex background is preferable.
Brief description
Fig. 1 is the corresponding flow chart of steps of extracting method of the present invention;
Fig. 2 is Fig. 1 corresponding steps flow chart instance graph;
Fig. 3 a is card image instance graph to be extracted of the present invention;
Fig. 3 b is the colour edging image rgbx instance graph of Fig. 3 a;
Fig. 3 c is the colour edging image rgby instance graph of Fig. 3 a;
Fig. 3 d is the colour edging image rgbimg instance graph of Fig. 3 a;
Fig. 4 is the edge image instance graph that S2 of the present invention is obtained using phase equalization algorithm;
Fig. 5 a is edge image img11 instance graph of the present invention;
Fig. 5 b is edge image img12 instance graph of the present invention;
Fig. 5 c is thick extraction image img1 instance graph of the present invention;
Fig. 6 is edge image img2 instance graph of the present invention;
Fig. 7 is final edge image img_z instance graph of the present invention;
Fig. 8 is obtained target card image instance graph by the present invention.
Specific embodiment
For making the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached in the embodiment of the present invention
Figure, is clearly and completely described to technical scheme it is clear that described embodiment is that a present invention part is real
Apply example, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not making creation
Property work under the premise of the every other embodiment that obtained, broadly fall into the scope of protection of the invention.
As Figure 1-Figure 2, the present invention devises one kind and is based on existing color image-adaptive edge detection method and phase
The card image extracting method of bit integrity detection method, comprises the following steps:
S1, the card image to be extracted to reading pre-process, so that image zooming-out improved efficiency, need not take very
Big memory storage space, is easy to real-time processing;Described pretreatment, it comprises the following steps:S11, the card to be extracted to reading
Image is filtered noise reduction process;S12, the image to filtered noise reduction process zoom in and out process.Specifically as this method
Preferred embodiment, described filtering noise reduction process includes adopting mean filter to process to complete the noise of aforementioned card original image is entered
Row suppression;The image down of filtered noise reduction process is the 1/2 of artwork size, with boosting algorithm efficiency and reduce memory storage
Space.
S2, the pretreated image of reading, are based respectively on color image-adaptive edge detection method, phase equalization inspection
Survey method, extracts each self-corresponding edge image from pretreated card image, and to two breadths currently being extracted
Edge image asked with computing after synthesis card image corresponding to thick extraction image;Specifically as the preferred embodiment of this method,
Corresponding edge is extracted from the pretreated card image as Fig. 3 a based on color image-adaptive edge detection method
The process of image includes:Rgb image corresponding to pretreated card image is carried out with horizontal edge detection to obtain through amount
The colour horizontal edge image rgbx changing, such as Fig. 3 b, and vertical rim detection is carried out to described rgb image card image, to obtain
Obtain quantified colored vertical edge image rgby, such as Fig. 3 c;Compare colored horizontal edge image rgbx, colored vertical edges one by one
The size of pixel value of edge image rgby corresponding pixel points simultaneously constructs coefficient matrix rgbmask, and this coefficient matrix rgbmask constructs
Principle is to judge that in colored horizontal edge image rgbx, whether the pixel value corresponding to current pixel point is more than colored vertical edge
Pixel value corresponding to current pixel point in image rgby, is that in coefficient matrix rgbmask, correspondence position puts 1, otherwise sets to 0;
Retain each layer of x of R, G, B triple channel of rgbx and rgby, the obvious edge in y direction, that is, the coefficient matrix obtaining is 1 point,
According to the following equation edge image extraction is carried out to colored horizontal edge image rgbx, colored vertical edge image rgby, to obtain
Obtain colour edging image rgbimg, such as Fig. 3 d;Described formula is
Rgbimg=rgbx rgbmask+rgby (1-rgbmask)
In above-mentioned formula, the dot product of matrix is exactly that each corresponding element of matrix is multiplied.
Specifically as the preferred embodiment of this method, based on phase equalization detection method, to pretreated card image
Carry out edge image extraction process process to include:Pretreated card image is switched to gray level image, and consistent based on phase place
Property detection algorithm obtain corresponding edge image, such as Fig. 4, because this algorithm is algorithms most in use, be therefore only applied to here
The principle background of the present invention briefly describes:So-called phase equalization, briefly just refers on each position of image each
A kind of metric form of the similarity of frequency content, it is a nondimensional amount, and its value is with the change of illumination and brightness no
Close, it can obtain the consistent position of phase height by searching for the peak value of local energy function, and that is, local energy function is being just
Ratio in phase equalization, after tested it may be concluded that the edge feature of image obtains in the phase spectrum uniformity of image
Embody well, the severe degree of the brightness of its testing result and image and contrast change is unrelated.Therefore, the method is in light
According to undesirable or image brightness distribution uneven in the case of can obtain preferable Detection results, read in original image,
And switched to gray level image, can get edge image using above-mentioned phase equalization algorithm.
Simultaneously because the colour edging image obtaining and the edge image that obtained using phase equalization method are inevitable
Some edges that have be pseudo-edge, so need remove pseudo-edge, to the two breadths edge currently being extracted in therefore described S2
Image asked with computing after the process of thick extraction image corresponding to synthesis card image comprise the steps:S21, respectively general
Two obtained width edge images switch to gray level image, determine each self-corresponding edge threshold using histogram;And pass through image
Morphological scale-space removes noise spot region (because grey value profile overwhelming majority value concentrates near smaller value, so we will
Front the 10% of gray value maximum is defined as edge, is then passed through morphological operation and removes less for image connected domain), with respectively
Obtain with the edge image img11 corresponding to color image-adaptive edge detection method and with phase equalization detection method
Corresponding edge image img12, such as Fig. 5 a, 5b,;S22, in order to remove pseudo-edge further it is believed that through two kinds of sides
The common first edges that method detects are real image edge, therefore obtained edge image img11, edge image img12 are entered
Row is asked and computing, to obtain initial edge image;And calculated really by described initial edge image is carried out with minimum external world rectangle
Recognize the object region corresponding to card extraction image, by entering the original image of object region and card to be extracted
Row dot product, obtains the thick extraction image of card image to be extracted, such as Fig. 5 c.
For the relative complex image of background, aforesaid operations also will detect that the edge of background texture, causes to extract mistake,
Therefore we take and carry out rim detection using phase equalization algorithm to thick result of extracting in HSV space, obtain final
Edge extraction result, said process is following step S3.
Specifically described S3, under HSV space, based on phase equalization detection method to described thick extraction image carry out side
Edge detects, to obtain final edge image;Specifically as the preferred embodiment of this method, described S3 comprises the steps:S31、
The thick extraction image reading in is transformed under hsv color space, and carries out next step behaviour using H component image and S component image
Make, because tone and saturation degree composition are closely related with the mode that people obtain color, and we solve in S2
The problem of brightness disproportionation, therefore do not consider V component image;S32, based on phase equalization detection method, the thick H extracting image is divided
Amount carries out rim detection, to obtain the edge strength image corresponding to H component and angular intensities image;And by the edge being obtained
After intensity image and angular intensities image are overlapped, carry out non-maximum restraining process, image skeletonization process obtains under H component
Edge image img21;S33, rim detection is carried out to the thick S component extracting image based on phase equalization detection method, with
Obtain the edge strength image corresponding to S component and angular intensities image;And will be strong to the edge strength being obtained image and angle
After degree image is overlapped, carry out non-maximum restraining process, the edge image obtaining under S component after image skeletonization process
img22;S34 is it is considered that the edge detecting through two methods is all real image edge, therefore to described edge image
Img21, edge image img22 are asked or computing obtains edge image img2, such as Fig. 6, then with thick extract image carry out asking with
Computing, obtains final edge image img_z, such as Fig. 7.
S4, based on edge image obtained in S3, the final extraction of card image is synthesized using Hough transformation detection of straight lines
Image.Detect the four edges of card using Hough transformation detection of straight lines, then calculate its minimum external world rectangle, through rotation and
Split and then obtain final card and extract image, because in hsv color space, color information is only had with H component and S component
Close, tone and saturation degree composition are closely related with the mode that people obtain color, and the method can effectively remove complex background line
Reason edge, the image robustness for complex background is preferable.
It is another object of the present invention to a kind of extraction system of card image will be provided it is characterised in that including:
1. pretreatment unit, this pretreatment unit can pre-process to the card image to be extracted reading in;Described pre-
Process, it includes being filtered noise reduction process to the card image to be extracted reading in;The image of filtered noise reduction process is carried out
Scaling is processed.Specifically as the preferred embodiment of this method, described filtering noise reduction process includes adopting mean filter to process to complete
The noise of aforementioned card original image is suppressed;The image down of filtered noise reduction process is the 1/2 of artwork size, with
Boosting algorithm efficiency simultaneously reduces memory storage space;
2. first order extraction unit, this first order extraction unit can based on color image-adaptive edge detection method,
Phase equalization detection method, extracts each self-corresponding edge image respectively from pretreated card image, and to work as
Front two extracted width edge images asked with computing after synthesis card image corresponding to thick extraction image;Concrete conduct is originally
The preferred embodiment of method, described is extracted from pretreated card image based on color image-adaptive edge detection method
The process of corresponding edge image includes:Level is carried out to the rgb image corresponding to the pretreated card image as Fig. 3 a
Rim detection to obtain quantified colour horizontal edge image rgbx, such as Fig. 3 b, and described rgb image card image is carried out
Vertically rim detection, to obtain quantified colored vertical edge image rgby, such as Fig. 3 c;Compare colored horizontal edge graph one by one
As rgbx, the size of the pixel value of colored vertical edge image rgby corresponding pixel points and construct coefficient matrix rgbmask, this is
Matrix number rgbmask structure principle is to judge in colored horizontal edge image rgbx the pixel value corresponding to current pixel point whether
More than the pixel value corresponding to current pixel point in colored vertical edge image rgby, it is then corresponding in coefficient matrix rgbmask
Position puts 1, otherwise sets to 0;Retain each layer of x of R, G, B triple channel of rgbx and rgby, the obvious edge in y direction, that is, obtain
Coefficient matrix is 1 point, carries out side to colored horizontal edge image rgbx, colored vertical edge image rgby according to the following equation
Edge image zooming-out, to obtain colour edging image rgbimg, such as Fig. 3 d;Described formula is
Rgbimg=rgbx rgbmask+rgby (1-rgbmask)
In above-mentioned formula, the dot product of matrix is exactly that each corresponding element of matrix is multiplied.
Specifically as the preferred embodiment of this method, based on phase equalization detection method, to pretreated card image
Carry out edge image extraction process process to include:Pretreated card image is switched to gray-scale map, and is based on phase equalization
Detection algorithm obtains corresponding edge image, such as Fig. 4, because this algorithm is algorithms most in use, is therefore only applied to this here
The principle background of invention briefly describes:So-called phase equalization, briefly just refers to each frequency on each position of image
A kind of metric form of the similarity of rate composition, it is a nondimensional amount, and its value is unrelated with the change of illumination and brightness
, it can obtain the consistent position of phase height by searching for the peak value of local energy function, i.e. local energy function direct ratio
In phase equalization, after tested it may be concluded that the edge feature of image has obtained very in the phase spectrum uniformity of image
Good embodiment, the severe degree of the brightness of its testing result and image and contrast change is unrelated.Therefore, the method is in illumination
Preferable Detection results can be obtained in the case that undesirable or image brightness distribution is uneven, read in original image, and
Switched to gray-scale map, be can get edge image using above-mentioned phase equalization algorithm;
Simultaneously because the colour edging image obtaining and the edge image that obtained using phase equalization method are inevitable
Some edges that have be pseudo-edge, so need remove pseudo-edge, can also complete in therefore described first order extraction unit
Following function:1st, respectively two obtained width edge images are switched to gray level image, determine each self-corresponding side using histogram
Edge threshold value;And by morphological image process remove noise spot region (due to grey value profile the overwhelming majority value concentrate on less
Near value, so we are defined as edge by before gray value maximum 10%, it is then passed through morphological operation image is less
Connected domain remove), with respectively obtain with the edge image img11 corresponding to color image-adaptive edge detection method and with
Edge image img12 corresponding to phase equalization detection method, such as Fig. 5 a, 5b;2nd, in order to remove pseudo-edge further, we
Think that the common first edges detecting through two methods are real image edge, therefore to obtained edge image img11, side
Edge image img12 is asked and computing, to obtain initial edge image;And by carrying out outside minimum to described initial edge image
Boundary rectangle calculates the object region confirming that card extracts corresponding to image, by by object region and card to be extracted
Original image carry out dot product, obtain the thick extraction image of card image to be extracted, such as Fig. 5 c.
3. edge image extraction unit, this edge image extraction unit can be under HSV space, based on phase equalization inspection
Survey method carries out rim detection to described thick extraction image, obtains final edge image;Specifically as the preferred reality of this method
Example, the processing procedure of described edge image extraction unit comprises the steps:First the thick extraction image reading in is transformed into HSV
Under color space, and carry out next step operation using H component image and S component image, because tone and saturation degree composition and people
Obtain color mode closely related, and we solve the problems, such as brightness disproportionation in S2, therefore do not consider V component
Image;Secondly based on phase equalization detection method, rim detection is carried out to the thick H component extracting image, to obtain H component institute
Corresponding edge strength image and angular intensities image;And the edge strength being obtained image and angular intensities image are folded
Plus after, carry out non-maximum restraining process, image skeletonization process obtains the edge image img21 under H component;It is again based on phase place
Consistency detecting method carries out rim detection to the thick S component extracting image, to obtain the edge strength image corresponding to S component
With angular intensities image;And after the edge strength being obtained image and angular intensities image are overlapped, carry out non-very big suppression
The edge image img22 under S component is obtained after system process, image skeletonization process;Finally it is believed that examining through two methods
The edge measured is all real image edge, therefore described edge image img21, edge image img22 is asked or computing
Obtain edge image img2, such as Fig. 6, then asked and computing with the thick image that extracts, obtain final edge image img_z, such as
Fig. 7.
4. second level extraction unit, this second level extraction unit can be based on the obtained edge graph of edge image extraction unit
Picture, synthesizes the final extraction image of card image, such as Fig. 8 using Hough transformation detection of straight lines.
The above, the only present invention preferably specific embodiment, but protection scope of the present invention is not limited thereto,
Any those familiar with the art the invention discloses technical scope in, technology according to the present invention scheme and its
Inventive concept equivalent or change in addition, all should be included within the scope of the present invention.
Claims (4)
1. a kind of extracting method of card image is it is characterised in that comprise the steps:
S1, the card image to be extracted to reading pre-process;
S2, it is based respectively on color image-adaptive edge detection method, phase equalization detection method, from pretreated card
Extract each self-corresponding edge image in image, and the two width edge images currently being extracted are asked with computing after synthesize
Thick extraction image corresponding to card image;
S3, under HSV space, based on phase equalization detection method to described thick extraction image carry out rim detection, with obtain
Final edge image;
S4, based on edge image obtained in S3, the final extraction image of card image is synthesized using Hough transformation detection of straight lines.
2. method according to claim 1 it is characterised in that:
In described S2, the two width edge images currently being extracted are asked with computing after synthesis card image corresponding to slightly carrying
The process taking image comprises the steps:
S21, two obtained width edge images are removed pseudo-edge process, that is, respectively by two obtained width edge images
Switch to gray level image, determine each self-corresponding edge threshold using histogram;And noise spot is removed by morphological image process
Region, with respectively obtain with the edge image img11 corresponding to color image-adaptive edge detection method and with phase place one
Edge image img12 corresponding to cause property detection method;
S22, obtained edge image img11, edge image img12 are asked and computing, to obtain initial edge image;
And calculated by described initial edge image is carried out with minimum external world rectangle, confirm that card extracts the target image corresponding to image
Region, by the original image of object region and card to be extracted is carried out dot product, obtains the thick of card image to be extracted
Extract image.
3. method according to claim 1 it is characterised in that:
Described S3 comprises the steps:
S31, by read in thick extraction image be transformed under hsv color space;
S32, rim detection is carried out to the thick H component extracting image based on phase equalization detection method, right to obtain H component institute
The edge strength image answered and angular intensities image;And the edge strength being obtained image and angular intensities image are overlapped
Afterwards, carry out non-maximum restraining process, image skeletonization process obtains the edge image img21 under H component;
S33, rim detection is carried out to the thick S component extracting image based on phase equalization detection method, right to obtain S component institute
The edge strength image answered and angular intensities image;And the edge strength being obtained image and angular intensities image are overlapped
Afterwards, non-maximum restraining process, the edge image img22 obtaining under S component after image skeletonization process are carried out;
S34, described edge image img21, edge image img22 are asked or computing, obtain after edge image img2 with thick
Extract image to be asked and computing, obtain final edge image img_z.
4. a kind of extraction system of card image is it is characterised in that include:
Pretreatment unit, this pretreatment unit can pre-process to the card image to be extracted reading in;
First order extraction unit, this first order extraction unit can be based on color image-adaptive edge detection method, phase place one
Cause property detection method, extracts each self-corresponding edge image respectively from pretreated card image, and to currently being carried
The two width edge images taking asked with computing after synthesis card image corresponding to thick extraction image;
Edge image extraction unit, this edge image extraction unit can be under HSV space, based on phase equalization detection method
Rim detection is carried out to described thick extraction image, obtains final edge image;
And second level extraction unit, this second level extraction unit can be based on the obtained edge graph of edge image extraction unit
Picture, synthesizes the final extraction image of card image using Hough transformation detection of straight lines.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610818707.2A CN106408533B (en) | 2016-09-12 | 2016-09-12 | A kind of card image extracting method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610818707.2A CN106408533B (en) | 2016-09-12 | 2016-09-12 | A kind of card image extracting method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106408533A true CN106408533A (en) | 2017-02-15 |
CN106408533B CN106408533B (en) | 2019-10-22 |
Family
ID=57999732
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610818707.2A Active CN106408533B (en) | 2016-09-12 | 2016-09-12 | A kind of card image extracting method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106408533B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107798325A (en) * | 2017-08-18 | 2018-03-13 | 中国银联股份有限公司 | Card identification method and equipment, computer-readable storage medium |
CN110287851A (en) * | 2019-06-20 | 2019-09-27 | 厦门市美亚柏科信息股份有限公司 | A kind of target image localization method, device, system and storage medium |
WO2021139169A1 (en) * | 2020-07-27 | 2021-07-15 | 平安科技(深圳)有限公司 | Method and apparatus for card recognition, device, and storage medium |
CN113989314A (en) * | 2021-10-26 | 2022-01-28 | 深圳前海环融联易信息科技服务有限公司 | Method for removing header and footer based on Hough transform linear detection |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104217444A (en) * | 2013-06-03 | 2014-12-17 | 支付宝(中国)网络技术有限公司 | Card area positioning method and equipment |
CN104268864A (en) * | 2014-09-18 | 2015-01-07 | 小米科技有限责任公司 | Card edge extracting method and device |
CN105095900A (en) * | 2014-05-04 | 2015-11-25 | 阿里巴巴集团控股有限公司 | Method and device of extracting specific information in standard card |
US20150371086A1 (en) * | 2013-06-30 | 2015-12-24 | Google Inc. | Extracting card data from multiple cards |
-
2016
- 2016-09-12 CN CN201610818707.2A patent/CN106408533B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104217444A (en) * | 2013-06-03 | 2014-12-17 | 支付宝(中国)网络技术有限公司 | Card area positioning method and equipment |
US20150371086A1 (en) * | 2013-06-30 | 2015-12-24 | Google Inc. | Extracting card data from multiple cards |
CN105095900A (en) * | 2014-05-04 | 2015-11-25 | 阿里巴巴集团控股有限公司 | Method and device of extracting specific information in standard card |
CN104268864A (en) * | 2014-09-18 | 2015-01-07 | 小米科技有限责任公司 | Card edge extracting method and device |
Non-Patent Citations (2)
Title |
---|
陈付梦等: ""基于提升小波的铁谱图像边缘检测"", 《机械科学与技术》 * |
马月娜等: ""基于边缘检测的人脸模拟画像检索"", 《计算机应用与软件》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107798325A (en) * | 2017-08-18 | 2018-03-13 | 中国银联股份有限公司 | Card identification method and equipment, computer-readable storage medium |
CN107798325B (en) * | 2017-08-18 | 2021-04-16 | 中国银联股份有限公司 | Card recognition method and apparatus, computer storage medium |
CN110287851A (en) * | 2019-06-20 | 2019-09-27 | 厦门市美亚柏科信息股份有限公司 | A kind of target image localization method, device, system and storage medium |
WO2021139169A1 (en) * | 2020-07-27 | 2021-07-15 | 平安科技(深圳)有限公司 | Method and apparatus for card recognition, device, and storage medium |
CN113989314A (en) * | 2021-10-26 | 2022-01-28 | 深圳前海环融联易信息科技服务有限公司 | Method for removing header and footer based on Hough transform linear detection |
Also Published As
Publication number | Publication date |
---|---|
CN106408533B (en) | 2019-10-22 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103136766B (en) | A kind of object conspicuousness detection method based on color contrast and color distribution | |
Yu et al. | Fast single image fog removal using edge-preserving smoothing | |
Chen et al. | A novel color edge detection algorithm in RGB color space | |
CN110264459A (en) | A kind of interstices of soil characteristics information extraction method | |
CN106408533A (en) | Card image extraction method and card image extraction system | |
CN109344724A (en) | A kind of certificate photo automatic background replacement method, system and server | |
CN111062931B (en) | Detection method of spliced and tampered image | |
CN106096491B (en) | Automatic identification method for microaneurysms in fundus color photographic image | |
CN103927741A (en) | SAR image synthesis method for enhancing target characteristics | |
CN107066952A (en) | A kind of method for detecting lane lines | |
Lo et al. | Joint trilateral filtering for depth map super-resolution | |
CN109544464A (en) | A kind of fire video image analysis method based on contours extract | |
Er-Sen et al. | An adaptive edge-detection method based on the canny operator | |
Guiming et al. | Remote sensing image edge-detection based on improved Canny operator | |
CN106446905A (en) | Surface crack texture extraction method based on fusion of seepage algorithm and adaptive Canny algorithm | |
CN112200742A (en) | Filtering and denoising method applied to edge detection | |
CN109389560A (en) | A kind of adaptive weighted filter image denoising method, device and image processing equipment | |
CN101493939A (en) | Method for detecting cooked image based on small wave domain homomorphic filtering | |
CN106228541A (en) | The method and device of screen location in vision-based detection | |
CN108280810A (en) | Automatic processing method for repairing cloud coverage area of single-time phase optical remote sensing image | |
CN106846271A (en) | A kind of method of reticulate pattern in removal identity card picture | |
CN106600615B (en) | A kind of Edge-Detection Algorithm evaluation system and method | |
CN110298835A (en) | A kind of leather surface damage testing method, system and relevant apparatus | |
CN111476723A (en) | Method for recovering lost pixels of remote sensing image with failed L andsat-7 scanning line corrector | |
CN112132848B (en) | Preprocessing method based on image layer segmentation and extraction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |