CN109961416B - Business license information extraction method based on morphological gradient multi-scale fusion - Google Patents
Business license information extraction method based on morphological gradient multi-scale fusion Download PDFInfo
- Publication number
- CN109961416B CN109961416B CN201910259552.7A CN201910259552A CN109961416B CN 109961416 B CN109961416 B CN 109961416B CN 201910259552 A CN201910259552 A CN 201910259552A CN 109961416 B CN109961416 B CN 109961416B
- Authority
- CN
- China
- Prior art keywords
- image
- pixel
- scale
- gray
- text
- 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.)
- Active
Links
- 230000000877 morphologic effect Effects 0.000 title claims abstract description 27
- 238000000605 extraction Methods 0.000 title claims abstract description 18
- 230000004927 fusion Effects 0.000 title claims abstract description 18
- 238000000034 method Methods 0.000 claims abstract description 41
- 238000012545 processing Methods 0.000 claims abstract description 32
- 238000001914 filtration Methods 0.000 claims abstract description 25
- 235000011036 Rubus leucodermis Nutrition 0.000 claims abstract description 19
- 235000003942 Rubus occidentalis Nutrition 0.000 claims abstract description 19
- 244000181917 Rubus leucodermis Species 0.000 claims abstract description 17
- 230000002146 bilateral effect Effects 0.000 claims abstract description 7
- 230000009466 transformation Effects 0.000 claims abstract description 5
- 230000005855 radiation Effects 0.000 claims description 13
- 238000001514 detection method Methods 0.000 claims description 10
- 238000010586 diagram Methods 0.000 claims description 8
- 238000009499 grossing Methods 0.000 claims description 7
- 230000007797 corrosion Effects 0.000 claims description 6
- 238000005260 corrosion Methods 0.000 claims description 6
- 230000003628 erosive effect Effects 0.000 claims description 6
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 claims description 3
- 230000010339 dilation Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 6
- 230000009286 beneficial effect Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 4
- 244000111388 Rubus occidentalis Species 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000005315 distribution function Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration by the use of local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- 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
- G06T2207/20024—Filtering details
- G06T2207/20028—Bilateral filtering
-
- 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
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30176—Document
Abstract
The invention discloses a business license information extraction method based on morphological gradient multi-scale fusion, which comprises the following steps: sequentially carrying out scaling processing of multi-scale transformation, denoising by a bilateral filtering algorithm and graying processing on the original image of the business license to obtain a gray image; carrying out black cap operation on the gray-scale image and inverting the gray-scale image according to pixels to obtain a black cap image; subtracting the gray-scale image and the black-hat image according to pixels to obtain an edge enhancement image; performing an opening operation on the edge enhancement image to obtain an opening operation image; detecting the open operation graph by using a maximum extremum stability algorithm to obtain a candidate region graph; detecting and de-duplicating the candidate region image to obtain text region images with different scales; and comparing the text region images under different scales, and reserving the text regions detected under each scale so as to extract image text information. The method solves the technical problem that the accuracy, hardware cost and calculation cost of the conventional image processing information in the aspect of business license information extraction cannot be considered at the same time, and greatly improves administrative efficiency.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a business license information extraction method based on morphological gradient multi-scale fusion.
Background
Images, as the visual basis of the world perceived by humans, are important means for humans to acquire, express and transmit information. Image processing techniques, which are techniques for analyzing and processing an image with a mathematical algorithm to achieve a desired result, generally include image compression, enhancement, restoration, description, and recognition. The image recognition technology is an important scientific research field, the development of the image recognition technology can greatly save the labor cost, for example, the character detection technology in the image recognition technology is used for document processing in the administrative aspect, so that the working pressure of the public staff can be effectively relieved, and the administrative efficiency can be improved.
The existing character detection technology is mainly divided into two types, one type is the traditional character detection technology based on a sliding window or a connected component and the like; another class is deep learning based text inspection techniques.
The traditional character detection technology based on the sliding window or the connected component is an image processing algorithm developed on the basis of geometry, the requirements on hardware in an engineering application environment are relatively low, the calculation cost is low, the accuracy rate is low, and the method is not suitable for information extraction of important files such as business license images.
While the character detection technology based on the deep learning algorithm, such as CTPN, SegLink, EAST, and the like, applies the neural network to the field of image processing, and although the accuracy of information extraction, such as character detection, is greatly improved, the requirements on hardware equipment in the engineering application environment are high and the calculation cost is huge because the neural node and the network thereof are too complex and a large amount of data is required for model training.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, provides a business license information extraction method based on morphological gradient multi-scale fusion, and solves the technical problem that the accuracy, hardware cost and calculation cost of the conventional image processing information extraction method cannot be considered at the same time.
In order to achieve the purpose of the invention, the invention adopts the technical scheme that: a business license information extraction method based on morphological gradient multi-scale fusion comprises the following steps:
s1, carrying out multi-scale transformation zooming processing on the original image of the business license to obtain N zooming images, wherein N is a positive integer, and the value is set according to the requirement of an engineering project;
s2, processing the zoom image by using a bilateral filtering algorithm to obtain a denoising smooth image;
s3, carrying out graying processing on the denoising smooth image to obtain a grayscale image;
s4, carrying out black cap operation on the gray level image and inverting according to pixels to obtain a black cap image;
s5, subtracting the gray-scale image and the black-hat image according to pixels to obtain an edge enhancement image;
s6, performing mathematical morphology open operation on the edge enhancement graph to obtain an open operation graph;
s7, performing text candidate region detection on the open operation graph by using a maximum stable extremum algorithm to obtain a candidate region graph;
s8, detecting the candidate region map, and removing repeated text candidate regions to obtain a text region map;
and S9, comparing the N text region maps under different scales, discarding the text regions detected only under specific or partial scales, reserving the text regions detected under all scales, obtaining the final text region map, and extracting the text information of the image.
The invention has the beneficial effects that: the mathematical morphology established on the basis of the lattice theory and the topology is used as the theoretical basis of the scheme of the invention, a set of image processing information extraction method with the steps being linked in a loop without consuming a large amount of hardware equipment resources is formed, and the accuracy of information extraction is ensured by the schemes of filtering, denoising, edge enhancement, multi-scale transformation and the like, so that the method can effectively and accurately extract useful information from the business license image.
Further, the method of the image scaling processing in step S1 is a bilinear interpolation scaling method.
The method has the following further beneficial effects: compared with a nearest neighbor interpolation method, the method has better zooming effect; compared with a bicubic interpolation method, the complexity is reduced, and hardware resources are saved.
Further, the bilateral filtering algorithm in step S2 includes the following steps:
s21, calculating the filtering weight by using the Euclidean distance of the pixel and the radiation difference in the pixel range domain, wherein the specific formula is as follows:
wherein, (i, j) and (k, l) are coordinates of two pixels, i and k are horizontal coordinates of the two pixels, and j and l are vertical coordinates of the two pixels; (i-k)2+(j-l)2The Euclidean distance of two pixel points; sigmadThe Euclidean distance smoothing parameter is obtained; i (I, j) and I (k, l) are pixel values of pixel points of coordinates (I, j) and (k, l), respectively; | | I (I, j) -I (k, l) | non-woven phosphor2Is the difference in radiation in the pixel domain; sigmarSmoothing parameters for the radiation difference; w (i, j, k, l) is the calculated filter weight.
S22, calculating the pixel value of the filtered image through the filtering weight and the pixel value of the zoom map, wherein the specific formula is as follows:
where I (k, l) is the pixel value of the scaled image; i. k is a pixel horizontal coordinate, and j and l are pixel vertical coordinates; i isD(i, j) are the filtered image pixel values.
The method has the following further beneficial effects: the Gaussian distribution function is used for carrying out weighted average operation on the two influence factors of the Euclidean distance of the pixel and the radiation difference to calculate the filtering weight value, so that two important factors of the Euclidean distance of the pixel and the radiation difference are considered simultaneously in the setting of the filtering weight value, the smooth filtering performance is better, and the denoising effect is excellent.
Further, the graying processing method in step S3 is to use three weights of 0.299, 0.587 and 0.114 to perform weighting processing on the R, G, B three components of the original color pixel, and sum up to obtain the grayscale pixel value, and the specific formula is as follows:
Gray(i，j)＝0.299*R(i，j)+0.587*G(i，j)+0.114*B(i，j)
wherein Gray (i, j) is the pixel value of the pixel point of the (i, j) coordinate; r (i, j), G (i, j), and B (i, j) are R component value, G component value, and B component value of the pixel point of the (i, j) coordinate, respectively.
The method has the following further beneficial effects: the R, G, B components representing the RGB intensities are weighted and summed rather than simply averaged together so that the pixel values effectively reflect the color intensity relationship of the color original.
Further, the black cap operation method in step S4 includes the steps of:
s41, firstly, the gray-scale image f and the morphological operation kernel b are subjected to mathematical expansion operation, and then the result and the morphological operation kernel b are subjected to mathematical corrosion operation, wherein the formula is as follows:
wherein f is1In order to obtain the result of the operation,
for a dilation operation, an erosion operation;
s42, combining the gray image f and the above operation result f1And performing subtraction operation to obtain a black cap image, wherein the specific formula is as follows: h ═ f-f1Wherein h is a black cap diagram.
The method has the following further beneficial effects: the image edge is smoothed, the isolated points are eliminated, and the adhesion of different objects in the image caused by the resolution is disconnected.
Further, the method of the on operation in step S6 is to enhance the edge of the image f2Firstly, performing mathematical corrosion operation with a morphological operation core b, and then performing mathematical expansion operation on the result, wherein the specific formula is as follows:
wherein f is3In order to operate on the pixel values of the image,
an expand operation, an erosion operation.
The method has the following further beneficial effects: the opening operation can remove isolated points and broken bands in the image processing process, so that the text structure is communicated into a text area.
Further, the step of detecting the candidate text region using the maximum stable extremum algorithm in step S7 is:
s71, performing binarization processing on the opening operation image by using pixel values of 0-255 as threshold values respectively to obtain 256 different binary images;
s72, comparing the areas of the connected regions in the 256 binary images, calculating the area change rate S of each connected region, and setting a reasonable area change rate threshold S according to the requirementthDetecting that the area change rate S is less than the threshold SthThe connected domain is the text candidate area.
The method has the following further beneficial effects: the binarization processing process is simple, and after the image is binarized, the pixel has only 255 and 0 values, which is beneficial to the discrimination of a connected region and the area calculation thereof; through simple comparison of the area change rates, the text candidate area can be judged, and the complexity is effectively reduced.
Drawings
Fig. 1 is a flowchart of a license information extraction method based on morphological gradient multi-scale fusion.
Fig. 2 is a zoom view of a license.
FIG. 3 is a denoised license smoothing diagram.
Fig. 4 is a gray scale image of a license.
Fig. 5 is a black cap picture of a business license.
Fig. 6 is a diagram of license edge enhancement.
Fig. 7 is a diagram illustrating a license operation.
Fig. 8 is a diagram of license candidates.
Fig. 9 is a diagram of regions of license text.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.
As shown in fig. 1, in an embodiment of the present invention, a method for extracting license information based on morphological gradient multi-scale fusion includes the following steps:
s1, in order to take the scaling effect and the hardware resource consumption into account, the business license original image is scaled by bilinear interpolation scaling method to obtain N scaling maps, where N is a positive integer, and the value is set according to the requirement of the engineering project, and the result is shown in fig. 2.
S2, processing the zoom image by using a bilateral filtering algorithm to obtain a de-noised smooth image, wherein the result is shown in FIG. 3;
the bilateral filtering algorithm is divided into two steps, firstly, the Euclidean distance of a pixel and the radiation difference in a pixel range domain are used for calculating the filtering weight; then calculating the pixel value of the filtered image through the filtering weight and the pixel value of the zoom image;
the specific formula for calculating the filtering weight is as follows:
wherein, (i, j) and (k, l) are coordinates of two pixels, i and k are horizontal coordinates of the two pixels, and j and l are vertical coordinates of the two pixels; (i-k)2+(j-l)2The Euclidean distance of two pixel points; sigmadThe Euclidean distance smoothing parameter is set according to engineering projects; i (I, j) and I (k, l) are pixel values of pixel points of coordinates (I, j) and (k, l), respectively; | | I (I, j) -I (k, l) | non-woven phosphor2Is the difference in radiation in the pixel domain; sigmarThe radiation difference smoothing parameter is set according to engineering projects; w (i, j, k, l) is the calculated filter weight.
The formula for calculating the filtered image pixel values by filtering the weights and scaling the pixel values of the map is:
where I (k, l) is the pixel value of the scaled image; i. k is a pixel horizontal coordinate, and j and l are pixel vertical coordinates; i isD(i, j) are the filtered image pixel values.
In the steps, the Gaussian distribution function is used for carrying out weighted average operation on the two influence factors of the Euclidean distance of the pixel and the radiation difference to calculate the filtering weight value, so that two important factors of the Euclidean distance of the pixel and the radiation difference are considered simultaneously in the setting of the filtering weight value, the smooth filtering performance is better, and the denoising effect is excellent.
S3, carrying out graying processing on the denoising smooth image to obtain a grayscale image, wherein the result is shown in FIG. 4; the graying processing method is that three weight values of 0.299, 0.587 and 0.114 are used for carrying out weighting processing on R, G, B three components of original color pixels, and the gray pixel values are obtained by summing, wherein the specific formula is as follows:
Gray(i，j)＝0.299*R(i，j)+0.587*G(i，j)+0.114*B(i，j)
wherein Gray (i, j) is the pixel value of the pixel point of the (i, j) coordinate; r (i, j), G (i, j), and B (i, j) are R component value, G component value, and B component value of the pixel point of the (i, j) coordinate, respectively.
The above steps add R, G, B components representing the RGB intensities in a weighted manner rather than simply adding them to average, so that the gray scale value can effectively reflect the color intensity relationship of the color original.
S4, performing black-cap operation on the gray-scale image and inverting the gray-scale image by pixel to obtain a black-cap image, wherein the result is shown in FIG. 5; the method for black cap operation is divided into two steps, firstly, the gray image f is firstly subjected to mathematical form expansion operation with a morphological operation core b, and then the result is subjected to mathematical form corrosion operation with the morphological operation core b, wherein the formula is as follows:
wherein f is1In order to obtain the result of the operation,for a dilation operation, an erosion operation;
then, the gray-scale image f is compared with the above-mentioned operation result f1And performing subtraction operation to obtain a black cap image, wherein the specific formula is as follows: h ═ f-f1Wherein h is a black cap diagram.
The black cap operation smoothes the image edge, eliminates isolated points, and breaks the adhesion of different objects in the image caused by the resolution.
S5, the gray-scale image and the black-hat image are subtracted by pixel to obtain an edge enhancement image for enhancing the text edge, and the result is shown in fig. 6.
S6, performing mathematical morphology opening operation on the edge enhancement map to obtain an opening operation map, the result of which is shown in fig. 7; the method of opening operation is to enhance the edge of the image f2Firstly, performing mathematical corrosion operation with a morphological operation core b, and then performing mathematical expansion operation on the result, wherein the specific formula is as follows:
wherein f is3To open the pixel values of the operation map,
an expand operation, an erosion operation.
The opening operation can remove isolated points and broken bands in the image processing process, so that the text structure is communicated into a text area.
S7, performing text candidate region detection on the open operation graph by using a maximum stable extremum algorithm to obtain a candidate region graph, wherein the result is shown in FIG. 8; the method comprises the following specific steps: firstly, respectively using pixel values of 0-255 as threshold value opening operation images to carry out binarization processing to obtain 256 different binary images; then comparing the areas of the connected regions in the 256 binary images, calculating the area change rate S of each connected region, and setting a reasonable area change rate threshold S according to the requirementthDetecting that the area change rate S is less than the thresholdValue SthThe connected domain is the text candidate area.
The binarization processing process is simple, and after the image is binarized, the pixel has only 255 and 0 values, which is beneficial to the discrimination of a connected region and the area calculation thereof; through simple comparison of the area change rates, the text candidate area can be judged, and the complexity is effectively reduced.
S8, detecting the candidate region map, removing repeated text candidate regions to obtain a text region map, wherein the result is shown in FIG. 9;
and S9, comparing the N text region maps under different scales, discarding the text regions detected only under specific or partial scales, reserving the text regions detected under all scales, obtaining the final text region map, and extracting the text information of the image.
The invention uses mathematical morphology established on the basis of lattice theory and topology as the theoretical basis of the scheme, forms a set of image processing information extraction method with the steps being linked with each other without consuming a large amount of hardware equipment resources, and ensures the accuracy of information extraction by using the schemes of filtering, denoising, edge enhancement, multi-scale transformation and the like, thereby effectively and accurately extracting useful information from the business license image. The method is not limited to processing business licenses, is also suitable for processing documents in other administrative aspects such as admission of identity card information, admission of student card information and the like, can effectively relieve the working pressure of public staff, and can improve the administrative efficiency.
Claims (10)
1. A business license information extraction method based on morphological gradient multi-scale fusion is characterized by comprising the following steps:
s1, carrying out scaling processing of multi-scale transformation on the original image of the business license to obtain N scaling images;
s2, processing the zoom image by using a bilateral filtering algorithm to obtain a denoising smooth image;
s3, carrying out graying processing on the denoising smooth image to obtain a grayscale image;
s4, carrying out black cap operation on the gray level image and inverting according to pixels to obtain a black cap image;
s5, subtracting the gray-scale image and the black-hat image according to pixels to obtain an edge enhancement image;
s6, performing mathematical morphology open operation on the edge enhancement graph to obtain an open operation graph;
s7, performing text candidate region detection on the open operation graph by using a maximum stable extremum algorithm to obtain a candidate region graph;
s8, detecting the candidate region map, and removing repeated text candidate regions to obtain a text region map;
and S9, comparing the N text region maps under different scales, reserving the text regions detected under each scale, obtaining the final text region map, and extracting the image text information.
2. The method for extracting license information based on morphological gradient multi-scale fusion as claimed in claim 1, wherein the method of image scaling in step S1 is bilinear interpolation scaling.
3. The method for extracting license information based on morphological gradient multi-scale fusion of claim 1, wherein the bilateral filtering algorithm in the step S2 comprises the following steps:
s21, calculating a filtering weight by using the Euclidean distance of the pixel and the radiation difference in the pixel range domain;
s22, calculating a filtered image pixel value by filtering the weight and scaling the pixel value of the map.
4. The method for extracting license information based on morphological gradient multi-scale fusion of claim 3, wherein the formula for calculating the filtering weight in step S21 is as follows:
wherein (i, j) and (k, l) are coordinates of two pixels,i. k is a two-pixel abscissa, and j and l are two-pixel ordinates; (i-k)2+(j-l)2The Euclidean distance of two pixel points; sigmadThe Euclidean distance smoothing parameter is obtained; i (I, j) and I (k, l) are pixel values of pixel points of coordinates (I, j) and (k, l), respectively; | | I (I, j) -I (k, l) | non-woven phosphor2Is the difference in radiation in the pixel domain; sigmarSmoothing parameters for the radiation difference; w (i, j, k, l) is the calculated filter weight.
5. The method for extracting license information based on morphological gradient multi-scale fusion as claimed in claim 4, wherein the formula for calculating the pixel value of the filtered image by filtering the weight and scaling the pixel value of the map in step S22 is as follows:
where I (k, l) is the pixel value of the scaled image; i. k is a pixel horizontal coordinate, and j and l are pixel vertical coordinates; i isD(i, j) are the filtered image pixel values.
6. The method for license information extraction based on morphological gradient multi-scale fusion of claim 1, wherein the graying in step S3 is performed by weighting R, G, B three components of original color pixels with three weights of 0.299, 0.587 and 0.114, and summing up to obtain gray pixel values.
7. The method for license information extraction based on morphological gradient multi-scale fusion as claimed in claim 6, wherein the formula for calculating the gray pixel value is:
Gray(i，j)＝0.299*R(i，j)+0.587*G(i，j)+0.114*B(i，j)
wherein Gray (i, j) is the pixel value of the pixel point of the (i, j) coordinate; r (i, j), G (i, j), and B (i, j) are R component value, G component value, and B component value of the pixel point of the (i, j) coordinate, respectively.
8. The method for extracting license information based on morphological gradient multi-scale fusion of claim 1, wherein the black cap operation method in step S4 comprises the following steps:
s41, firstly, the gray-scale image f and the morphological operation kernel b are subjected to mathematical expansion operation, and then the result and the morphological operation kernel b are subjected to mathematical corrosion operation, wherein the formula is as follows:
wherein f is1In order to obtain the result of the operation,
for a dilation operation, an erosion operation;
s42, combining the gray image f and the above operation result f1And performing subtraction operation to obtain a black cap image, wherein the specific formula is as follows: h ═ f-f1Wherein h is a black cap diagram.
9. The method for extracting license information based on morphological gradient multi-scale fusion of claim 1, wherein the method of the opening operation in step S6 is to apply an edge enhancement map f2Firstly, performing mathematical corrosion operation with a morphological operation core b, and then performing mathematical expansion operation on the result, wherein the specific formula is as follows:
wherein f is3In order to operate on the pixel values of the image,
an expand operation, an erosion operation.
10. The method for extracting business license information based on morphological gradient multi-scale fusion of claim 1, wherein the step of text candidate region detection using maximum stable extremum algorithm in step S7 is as follows:
s71, performing binarization processing on the opening operation image by using pixel values of 0-255 as threshold values respectively to obtain 256 different binary images;
s72, comparing the areas of the connected regions in the 256 binary images, calculating the area change rate S of each connected region, and setting a reasonable area change rate threshold S according to the requirementthDetecting that the area change rate S is less than the threshold SthThe connected domain is the text candidate area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910259552.7A CN109961416B (en) | 2019-04-02 | 2019-04-02 | Business license information extraction method based on morphological gradient multi-scale fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910259552.7A CN109961416B (en) | 2019-04-02 | 2019-04-02 | Business license information extraction method based on morphological gradient multi-scale fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109961416A CN109961416A (en) | 2019-07-02 |
| CN109961416B true CN109961416B (en) | 2020-11-10 |
Family
ID=67025580
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910259552.7A Active CN109961416B (en) | 2019-04-02 | 2019-04-02 | Business license information extraction method based on morphological gradient multi-scale fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109961416B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110379178B (en) * | 2019-07-25 | 2021-11-02 | 电子科技大学 | Intelligent unmanned automobile parking method based on millimeter wave radar imaging |
| CN112749704A (en) * | 2019-10-31 | 2021-05-04 | 北京金山云网络技术有限公司 | Text region detection method and device and server |
| CN112785508A (en) * | 2019-11-11 | 2021-05-11 | 珠海金山办公软件有限公司 | Method and device for denoising electronic document picture |
| CN112330553B (en) * | 2020-10-30 | 2022-07-01 | 武汉理工大学 | Crack image denoising method, device and storage medium |
| CN115082485B (en) * | 2022-08-23 | 2023-08-29 | 广东欧达雅包装制品有限公司 | Method and system for detecting bubble defects on surface of injection molding product |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9014480B2 (en) * | 2012-07-19 | 2015-04-21 | Qualcomm Incorporated | Identifying a maximally stable extremal region (MSER) in an image by skipping comparison of pixels in the region |
| CN108399405B (en) * | 2017-02-07 | 2023-06-27 | 腾讯科技（上海）有限公司 | Business license identification method and device |
| CN108898138A (en) * | 2018-05-30 | 2018-11-27 | 西安理工大学 | Scene text recognition methods based on deep learning |
-
2019
- 2019-04-02 CN CN201910259552.7A patent/CN109961416B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109961416A (en) | 2019-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109961416B (en) | Business license information extraction method based on morphological gradient multi-scale fusion | |
| Xie et al. | Image edge detection based on opencv | |
| CN102782706B (en) | Text enhancement of a textual image undergoing optical character recognition | |
| CN110033471B (en) | Frame line detection method based on connected domain analysis and morphological operation | |
| CN108280823A (en) | The detection method and system of the weak edge faults of cable surface in a kind of industrial production | |
| CN108564597B (en) | Video foreground object extraction method fusing Gaussian mixture model and H-S optical flow method | |
| CN104574328A (en) | Color image enhancement method based on histogram segmentation | |
| CN110428450B (en) | Scale-adaptive target tracking method applied to mine tunnel mobile inspection image | |
| CN113111878B (en) | Infrared weak and small target detection method under complex background | |
| CN115272306B (en) | Solar cell panel grid line enhancement method utilizing gradient operation | |
| CN102930508A (en) | Image residual signal based non-local mean value image de-noising method | |
| CN111027637A (en) | Character detection method and computer readable storage medium | |
| CN107358635B (en) | Color morphological image processing method based on fuzzy similarity | |
| Gupta et al. | Fast single image haze removal method for inhomogeneous environment using variable scattering coefficient | |
| CN111179289B (en) | Image segmentation method suitable for webpage length graph and width graph | |
| Haller | Adaptive histogram equalization in GIS | |
| Li et al. | A study of crack detection algorithm | |
| CN112288726A (en) | Method for detecting foreign matters on belt surface of underground belt conveyor | |
| CN105354823A (en) | Tree-ring image edge extraction and segmentation system | |
| CN111462084B (en) | Image vectorization printing bleeding point prediction system and method based on random forest | |
| CN114529715A (en) | Image identification method and system based on edge extraction | |
| Malhotra et al. | Image denoising with various filtering techniques | |
| CN110766624B (en) | Point target and dark spot image background balancing method based on iterative restoration | |
| Zhang et al. | Using Gaussian Kernels to Remove Uneven Shading from a Document Image | |
| CN110782405B (en) | Point target and dark spot image background balancing method based on gradient identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |