CN106780535A - A kind of gray level image processing method - Google Patents
A kind of gray level image processing method Download PDFInfo
- Publication number
- CN106780535A CN106780535A CN201611188925.9A CN201611188925A CN106780535A CN 106780535 A CN106780535 A CN 106780535A CN 201611188925 A CN201611188925 A CN 201611188925A CN 106780535 A CN106780535 A CN 106780535A
- Authority
- CN
- China
- Prior art keywords
- image
- gray
- pixel
- text block
- gray level
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 8
- 238000000034 method Methods 0.000 claims description 19
- 238000004458 analytical method Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000009499 grossing Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 3
- 230000000750 progressive effect Effects 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- 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
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Character Input (AREA)
Abstract
The present invention relates to a kind of gray level image processing method, comprising four steps:The threshold value of gray image is calculated, the textural characteristics of gray image is protruded, the character in the text block split in gray level image, identification text block, useful achievement of the invention is:The calculating that the present invention passes through threshold value, gray level image is split, then, textural characteristics to gray image are further protruded, character therein and word are recognized by determining the border of text block and splitting text block, the present invention considers comprehensive, progressive, not only allows for matching character in text block, while also to match the word in text block.
Description
Technical field
The present invention relates to image processing field, it is related to a kind of gray level image processing method.
Background technology
In recent years, with the popularization and internet development of portable equipment for taking photograph equipment, the quantity of digital picture and digital video
Just increase with surprising rapidity.And gray level image is wherein conventional storage mode, coloured image, video are usually with gray level image
As the intermediate form of conversion.Embedded word in the picture is the important expression way of image, semantic.If can be using meter
Calculation machine is automatically positioned and recognizes these words, it is possible to allow the content of computer automatic understanding image, and then by ripe
Text retrieval technique retrieve image, so as to providing a kind of approach for the image procossing based on content.
The content of the invention
In view of this, the present invention provides a kind of gray level image processing method for solving or partly solving the above problems.
To reach the effect of above-mentioned technical proposal, the technical scheme is that:A kind of gray level image processing method, comprising
Following steps:
1) threshold value of gray level image is calculated, one is selected by the histogrammic discriminant analysis of grey level to gray level image
Individual global threshold, splits according to global threshold to gray level image, and the pixel point value less than threshold value is set into black, is more than
The pixel of threshold value is set to white, the gray level image after being processed;
2) pixel of the gray image after treatment is divided into foreground pixel or background pixel according to gray value, so as to process
Gray image afterwards is also divided into foreground image and background image two parts, for foreground image and background image in it is each
Individual pixel p, calculates one group of new threshold value F=(Xmax+Xmin)/4, wherein, XmaxIt is maximum in the neighborhood 1cm*1cm of pixel p
Gray value, XminIt is gray value minimum in the neighborhood 1cm*1cm of pixel p, F represents new threshold value, new for each
Threshold value, combined with texture feature re-starts one group of texture threshold of calculating, selection one from grey level's histogram of gray level image
Immediate with the value of corresponding new threshold value as optimal threshold value in group texture threshold, the texture as gray image is special
Levy;
3) textural characteristics of gray image are brought into, the side of foreground image and background image is extracted using Laplace operator
Boundary, the connection of row bound is entered using smoothing algorithm, carries out projected outline's analysis to determine the border of text block, by text block one
It is individual to be proposed from gray-scale map, reliable stroke is extracted from text block as benchmark, calculate Gauss model parameter and carry out rough segmentation
Cut, based on color homogeneity distribution with connect body method and further filter noise therein, so as to obtain the text split
Block;
4) character therein is recognized using nearest neighbor classifier from the text block split, sets up dictionary and character library,
Character library is generated into template image, is matched with template image using the character in the text block that SIFT feature will have been split, then
Matching result is modified using geometric verification algorithm, matching result is contrasted one by one with the word in dictionary, obtained
Word in the text block split for meeting rule in dictionary.
Useful achievement of the invention is:The calculating that the present invention passes through threshold value, splits, then, to ash to gray level image
The textural characteristics of color image are further protruded, by determine text block border and split text block come recognize character therein with
And word, it is of the invention to consider comprehensive, progressive, not only allow for matching character in text block, while also will be to text
Word in block is matched.
Specific embodiment
In order that the technical problems to be solved by the invention, technical scheme and beneficial effect become more apparent, below tie
Embodiment is closed, the present invention will be described in detail.It should be noted that specific embodiment described herein is only used to explain
The present invention, is not intended to limit the present invention, and can realize that the product of said function belongs to equivalent and improvement, is all contained in this hair
Within bright protection domain.Specific method is as follows:
Embodiment 1:Automatic image procossing is a processing procedure comprising many steps.In those steps, it is most heavy
The link wanted is exactly that image is pre-processed, and image is converted into binary image.The binary conversion treatment of image, being exactly will figure
As the gray value of each upper pixel is set to 0 or 255, that is, original coloured image is converted into whole image it is in
Reveal the gray level image for significantly there was only black and white visual effect.For the image containing word, its purpose is typically
Separated word segment as prospect, effect directly influences the effect of follow-up text segmentation and recognition.
A kind of gray level image processing method, comprises the steps of:
1) threshold value of gray level image is calculated, one is selected by the histogrammic discriminant analysis of grey level to gray level image
Individual global threshold, splits according to global threshold to gray level image, and the pixel point value less than threshold value is set into black, is more than
The pixel of threshold value is set to white, the gray level image after being processed;
2) pixel of the gray image after treatment is divided into foreground pixel or background pixel according to gray value, so as to process
Gray image afterwards is also divided into foreground image and background image two parts, for foreground image and background image in it is each
Individual pixel p (x, y), calculates one group of new threshold value F (x, y)=(Xmax+Xmin)/4, wherein, XmaxIt is the neighbour of pixel p (x, y)
Maximum gray value, X in the 1cm*1cm of domainminIt is gray value minimum in the neighborhood 1cm*1cm of pixel p (x, y), for each
Individual new threshold value, combined with texture feature re-starts one group of texture threshold of calculating from grey level's histogram of gray level image,
Select immediate with the value of corresponding new threshold value as optimal threshold value in one group of texture threshold, as the line of gray image
Reason feature;
3) textural characteristics of gray image are brought into, the side of foreground image and background image is extracted using Laplace operator
Boundary, the connection of row bound is entered using smoothing algorithm, carries out projected outline's analysis to determine the border of text block, by text block one
It is individual to be proposed from gray-scale map, reliable stroke is extracted from text block as benchmark, calculate Gauss model parameter and carry out rough segmentation
Cut, based on color homogeneity distribution with connect body method and further filter noise therein, so as to obtain the text split
Block;
4) character therein is recognized using nearest neighbor classifier from the text block split, sets up dictionary and character library,
Character library is generated into template image, is matched with template image using the character in the text block that SIFT feature will have been split, then
Matching result is modified using geometric verification algorithm, matching result is contrasted one by one with the word in dictionary, obtained
Word in the text block split for meeting rule in dictionary.
Embodiment 2:Text segmentation is from the text filed middle process for separating text pixel point.
Edge feature is that image procossing uses more early, more feature, and rim detection is to be mutated to divide based on pixel grey scale
Cut the most common method of image.Method based on edge feature assume that have between word and background very strong contrast and
Text edges are clear, and text filed detection is carried out according to text filed abundant marginal information.This class algorithm generally first makes
Border is obtained with boundary extraction algorithm (such as operator, operator, operator) etc.;Reusing smoothing algorithm or morphological method is carried out
Contour connection, thus obtains complete word border.The maximum restraining factors of method based on edge are to work as background
Profile and text profile intersect adhesion when, word and background will by it is mixed it is quiet together with processed, finally
Background area may be taken as to remove, or obtain one and include word and the region comprising background.Therefore others side is needed
Method is verified.
By the observation to a large amount of Chinese characters and English alphabet, angle point in word is by substantial amounts of straightway and few for the present embodiment
The curved section of amount links together composition, particularly Chinese character.It is therefore contemplated that there is substantial amounts of angle point in image
Region be that text filed probability is very big.And the number of angle point is few or region of without angle point is background area in image
The probability in domain is very big.Where can to a certain extent judging the curve with straight line long and smaller curvature using angle point information
Region is not this paper regions.
Texture is the concept commonly used in image analysis processing, and it is used to describe the local characteristicses of image.Although at present still
So think " as surface or the texture of image attributes, without the definition that can be widely accepted ", but in general, texture refers to
The repeat pattern of image local change, certain deterministic rule by pixel in its neighborhood space or statistical rule come
The local random and macroscopical regular characteristic of description.
Denoising is carried out to gray level image using Gauss denoising method, Gauss denoising method is to all pixels point in image
Gray value be weighted the process that is averaged after summation, the entirely gray value of all pixels point, its own and neighborhood
Then the gray value of other interior pixels is averaged by weighted sum and obtained.The specific algorithm process of Gauss denoising method
It is:Slided in entire image with a window (or convolution, template), with the weighted average gray value of all pixels in window
Go the value of replacement window center pixel.
The preferred embodiments of the invention is the foregoing is only, claims of the invention are not limited to.
Simultaneously it is described above, for those skilled in the technology concerned it would be appreciated that and implement, therefore other be based on institute of the present invention
The equivalent change that disclosure is completed, should be included in the covering scope of the claims.
Useful achievement of the invention is:The calculating that the present invention passes through threshold value, splits, then, to ash to gray level image
The textural characteristics of color image are further protruded, by determine text block border and split text block come recognize character therein with
And word, it is of the invention to consider comprehensive, progressive, not only allow for matching character in text block, while also will be to text
Word in block is matched.
Claims (1)
1. a kind of gray level image processing method, it is characterised in that comprise the steps of:
1) threshold value of gray level image is calculated, one is selected by the histogrammic discriminant analysis of grey level to gray level image entirely
Office's threshold value, splits according to the global threshold to gray level image, and the pixel point value less than the global threshold is set to
Black, the pixel more than the global threshold is set to white, the gray level image after being processed;
2) pixel of the gray image after the treatment is divided into foreground pixel or background pixel according to gray value, so that described
Gray image after treatment is also divided into foreground image and background image two parts, for the foreground image and the background
Each pixel p in image, calculates one group of new threshold value F=(Xmax+Xmin)/4, wherein, XmaxIt is the neighborhood of pixel p
Maximum gray value, X in 1cm*1cmminIt is gray value minimum in the neighborhood 1cm*1cm of pixel p, F represents the new threshold
Value, for threshold value new each described, combined with texture feature is entered again from grey level's histogram of the gray level image
Row calculates one group of texture threshold, the immediate conduct of value in texture threshold described in one group of selection with the corresponding new threshold value
Optimal threshold value, as the textural characteristics of gray image;
3) textural characteristics of the gray image are brought into, the foreground image and the background is extracted using Laplace operator
The border of image, the connection of row bound is entered using smoothing algorithm, carries out projected outline's analysis to determine gray level image Chinese version block
Border, by the text block one by one from gray-scale map propose, reliable stroke is extracted from the text block as benchmark,
Calculating Gauss model parameter simultaneously carries out coarse segmentation, based on color homogeneity distribution with connect body method and further filter therein making an uproar
Sound, so as to obtain the text block split;
4) character therein is recognized using nearest neighbor classifier from the text block split, sets up dictionary and word
Storehouse, template image is generated by character library, using SIFT feature by the character and the template in the text block split
Image is matched, and reuses geometric verification algorithm to be modified matching result, in the matching result and dictionary
Word contrasted one by one, be met the word in the good text block of the segmentation of rule in dictionary.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611188925.9A CN106780535A (en) | 2016-12-21 | 2016-12-21 | A kind of gray level image processing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611188925.9A CN106780535A (en) | 2016-12-21 | 2016-12-21 | A kind of gray level image processing method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106780535A true CN106780535A (en) | 2017-05-31 |
Family
ID=58893438
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611188925.9A Pending CN106780535A (en) | 2016-12-21 | 2016-12-21 | A kind of gray level image processing method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106780535A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107292305A (en) * | 2017-06-20 | 2017-10-24 | 北京小米移动软件有限公司 | Character rotation method and device |
CN110533040A (en) * | 2019-09-05 | 2019-12-03 | 哈尔滨理工大学 | A kind of annular region binary image processing method and processing device |
CN111161247A (en) * | 2019-12-30 | 2020-05-15 | 凌云光技术集团有限责任公司 | Detection method for variable code reading character quality verification |
CN115100509A (en) * | 2022-07-15 | 2022-09-23 | 山东建筑大学 | Image identification method and system based on multi-branch block-level attention enhancement network |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1790377A (en) * | 2004-12-17 | 2006-06-21 | 佳能株式会社 | Reverse character recognition method, quick and accurate block sorting method and text line generation method |
CN102930277A (en) * | 2012-09-19 | 2013-02-13 | 上海珍岛信息技术有限公司 | Character picture verification code identifying method based on identification feedback |
CN102968637A (en) * | 2012-12-20 | 2013-03-13 | 山东科技大学 | Complicated background image and character division method |
CN105373794A (en) * | 2015-12-14 | 2016-03-02 | 河北工业大学 | Vehicle license plate recognition method |
CN105894037A (en) * | 2016-04-21 | 2016-08-24 | 北京航空航天大学 | Whole supervision and classification method of remote sensing images extracted based on SIFT training samples |
-
2016
- 2016-12-21 CN CN201611188925.9A patent/CN106780535A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1790377A (en) * | 2004-12-17 | 2006-06-21 | 佳能株式会社 | Reverse character recognition method, quick and accurate block sorting method and text line generation method |
CN102930277A (en) * | 2012-09-19 | 2013-02-13 | 上海珍岛信息技术有限公司 | Character picture verification code identifying method based on identification feedback |
CN102968637A (en) * | 2012-12-20 | 2013-03-13 | 山东科技大学 | Complicated background image and character division method |
CN105373794A (en) * | 2015-12-14 | 2016-03-02 | 河北工业大学 | Vehicle license plate recognition method |
CN105894037A (en) * | 2016-04-21 | 2016-08-24 | 北京航空航天大学 | Whole supervision and classification method of remote sensing images extracted based on SIFT training samples |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107292305A (en) * | 2017-06-20 | 2017-10-24 | 北京小米移动软件有限公司 | Character rotation method and device |
CN110533040A (en) * | 2019-09-05 | 2019-12-03 | 哈尔滨理工大学 | A kind of annular region binary image processing method and processing device |
CN111161247A (en) * | 2019-12-30 | 2020-05-15 | 凌云光技术集团有限责任公司 | Detection method for variable code reading character quality verification |
CN111161247B (en) * | 2019-12-30 | 2023-10-20 | 凌云光技术股份有限公司 | Detection method for variable code reading character quality verification |
CN115100509A (en) * | 2022-07-15 | 2022-09-23 | 山东建筑大学 | Image identification method and system based on multi-branch block-level attention enhancement network |
CN115100509B (en) * | 2022-07-15 | 2022-11-29 | 山东建筑大学 | Image identification method and system based on multi-branch block-level attention enhancement network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107609549B (en) | Text detection method for certificate image in natural scene | |
CN104751142B (en) | A kind of natural scene Method for text detection based on stroke feature | |
Poletti et al. | A review of thresholding strategies applied to human chromosome segmentation | |
CN110717896B (en) | Plate strip steel surface defect detection method based on significance tag information propagation model | |
CN103886589B (en) | Object-oriented automated high-precision edge extracting method | |
CN110197153B (en) | Automatic wall identification method in house type graph | |
CN106780535A (en) | A kind of gray level image processing method | |
CN104680127A (en) | Gesture identification method and gesture identification system | |
CN109241973B (en) | Full-automatic soft segmentation method for characters under texture background | |
CN103186904A (en) | Method and device for extracting picture contours | |
CN106127817B (en) | A kind of image binaryzation method based on channel | |
WO2015092904A1 (en) | Image-processing apparatus, image-processing method, and image-processing program | |
CN104766316B (en) | New lip partitioning algorithm in tcm inspection | |
WO2020140198A1 (en) | Fingernail image segmentation method, apparatus and device, and storage medium | |
CN110309830B (en) | Carapace bone word automatic segmentation method based on mathematical morphology and regional connectivity | |
CN114863492B (en) | Method and device for repairing low-quality fingerprint image | |
CN103049756A (en) | Method for automatically extracting and removing words in color image on basis of CEMA (Cellular Message Encryption Algorithm) and texture matching repairing technology | |
CN114863493A (en) | Detection method and detection device for low-quality fingerprint image and non-fingerprint image | |
CN112686265A (en) | Hierarchic contour extraction-based pictograph segmentation method | |
CN103971367A (en) | Hydrologic data image segmenting method | |
CN104268845A (en) | Self-adaptive double local reinforcement method of extreme-value temperature difference short wave infrared image | |
CN107038401A (en) | Lip contour segmentation and feature extraction method | |
JP6132485B2 (en) | Image processing apparatus, operation method of image processing apparatus, and image processing program | |
CN108205678A (en) | A kind of nameplate Text region processing method containing speck interference | |
CN104867114B (en) | Blind minimizing technology is permeated at a kind of ancient books scan image back side based on condition random field |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170531 |