CN109636743A - A kind of method and device removing picture noise - Google Patents
A kind of method and device removing picture noise Download PDFInfo
- Publication number
- CN109636743A CN109636743A CN201811415930.8A CN201811415930A CN109636743A CN 109636743 A CN109636743 A CN 109636743A CN 201811415930 A CN201811415930 A CN 201811415930A CN 109636743 A CN109636743 A CN 109636743A
- Authority
- CN
- China
- Prior art keywords
- region
- image
- channel
- noise
- connected region
- 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
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000002372 labelling Methods 0.000 claims abstract description 9
- 238000001914 filtration Methods 0.000 claims description 3
- 230000004069 differentiation Effects 0.000 claims 1
- 239000003550 marker Substances 0.000 claims 1
- 230000008901 benefit Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
Classifications
-
- G06T5/70—
-
- G06T5/73—
-
- G06T5/94—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
Abstract
The invention discloses a kind of methods for removing picture noise, include the following steps: S01: obtaining image to be processed, described image is decomposed on RGB color domain;S02: the first-order difference characterisitic function g (x, y) of different channel images is calculated separately;S03: the binary map function B (x, y) of first-order difference characterisitic function g (x, y) described in each channel image is obtained;S04: region binary map function B (x, y) in each channel image is marked using connected region labeling algorithm, forms connected region;S05: being labeled as noise region for the connected region that area in each channel image is less than this channel corresponding region threshold value, and the connected region that area is greater than this channel corresponding region threshold value is labeled as fringe region;S06: noise region described in each channel image and fringe region are respectively processed.A kind of method and device removing picture noise provided by the invention, recognizes noise region and fringe region according to the connection characteristic at noise and edge, can Protect edge information region while suppressing noise.
Description
Technical field
The present invention relates to field of data recognition, and in particular to a kind of method and device for removing picture noise.
Background technique
Since there are various electrical noises, institutes in imaging process for imaging sensor (CIS, CMOS Image Sensor)
To be presented as that various noises, tradition model noise attribute by various statistical models in image on piece, but in reality
In the image processing process of border, noise region generally requires to trade off with fringe region, i.e., to noise suppressed while, marginal zone
Domain can be also smoothed.
The existing this model for not distinguishing fringe region and noise region and processing method when handling noise,
So that fringe region is also destroyed, so that the fringe region of whole image is imperfect, and then the matter of whole image is influenced
Amount.
Summary of the invention
Technical problem to be solved by the invention is to provide it is a kind of remove picture noise method and device, according to noise with
The connection characteristic at edge recognizes noise region and fringe region, can while inhibiting noise region Protect edge information area
Domain.
To achieve the goals above, the present invention adopts the following technical scheme: a kind of method for removing picture noise, including such as
Lower step:
S01: obtaining image to be processed, and described image is decomposed on RGB color domain, forms R channel image, and G is logical
Road image and channel B image;
S02: the first-order difference characterisitic function g (x, y) of different channel images is calculated separately;
S03: obtain the corresponding binary map function B of first-order difference characterisitic function g (x, y) described in each channel image (x,
y);The binary map function B (x, y)=g (x, y) > thr;The thr indicates differential threshold;
S04: marking the corresponding region binary map function B (x, y) in each channel image using connected region labeling algorithm,
Connected region is formed, and area statistics are carried out to each connected region;
S05: being labeled as noise region for the connected region that area in each channel image is less than this channel corresponding region threshold value,
The connected region that area is greater than this channel corresponding region threshold value is labeled as fringe region;
S06: being smoothed noise region described in each channel image, is sharpened place to the fringe region
Each channel image after processing is synthesized new image, the as image after removal noise by reason.
Further, the first-order difference characterisitic function g (x, y)=‖ f (x-1, y)-f (x+1, y) ‖+‖ f (x, y-1)-f
(x,y+1)‖;Wherein, f (x, y) indicates original image, | | * | | it is norm or gradient operator.
Further, the connected region labeling algorithm is 8 connections or 4 connections.
Further, described includes counting to the pixel total amount of connected region to each connected region progress area statistics.
Further, the step S02 includes: the first-order difference characterisitic function g for calculating described image R channel image1(x,
y);Calculate the first-order difference characterisitic function g of described image channel B image2(x,y);Calculate the single order of described image G channel image
Differential characteristic function g3(x,y)。
Further, the step S03 includes: the binary map function B for calculating R channel image1(x,y);Calculate channel B figure
The binary map function B of picture2(x,y);Calculate the binary map function B of G channel image3(x,y)。
Further, the corresponding region of binary map function in R channel image, channel B image and G channel image is marked respectively
It is denoted as m, the region for being marked as m in different channel images is marked using connected region labeling algorithm, forms x connected region, wherein
X is the integer more than or equal to 1;
Further, the noise region is smoothed using smooth filtering method.
A kind of device removing picture noise provided by the invention, including the judgement of first-order difference generation module, first-order difference
Module, connected region mark module, connected region area statistics module, discrimination module, noise processed module and edge processing module;To
It handles image to input in the first-order difference generation module, forms the first-order difference characterisitic function g (x, y) of subchannel image, and
It is transmitted to the first-order difference judgment module, the first-order difference judgment module generates the first-order difference characterisitic function g (x, y)
Binary map function B (x, y), and be transmitted to the connected region mark module, the connected region mark module is according to binary map letter
Number B (x, y) is marked, and forms the connected region of each channel image, and be transmitted to the connected region area statistics module, described
Connected region area statistics module counts the area of each connected region, and statistical result is transmitted to the discrimination module, described to sentence
Other module differentiates that each connected region is noise region or fringe region according to the corresponding region threshold in each channel, and will judge
Noise region out is transmitted to the noise processed module and is handled, and the fringe region judged is transmitted to the side
Edge processing module is handled, by each channel image composograph after processing, the as image after removal noise.
The invention has the benefit that the present invention is counted using connected region, it can effectively judge sudden change region for noise region
Or fringe region, can when handling noise region effective protection image edge area, can be further to noise
Region and fringe region carry out different disposal, while handling the amendment of noise region and fringe region well;Using subchannel image
The method for handling noise, can efficiently and rapidly remove the noise of channel image unless each, further increase overall picture quality.
Detailed description of the invention
Attached drawing 1 is a kind of flow chart for removing picture noise method of the present invention.
Attached drawing 2 is a kind of structural schematic diagram for removing picture noise device of the present invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, with reference to the accompanying drawing to specific reality of the invention
The mode of applying is described in further detail.
The maximum difference at noise and edge be noise be it is isolated, edge is connected on a large scale.Come from image statistics
It sees, edge span is generally the belt-like zone that multiple pixels are constituted.So a scale of image can be calculated by various operators
It is intrinsic, the distribution of fringe region and noise region is obtained, the connected region area in statistics first-order difference figure is then passed through.Face to face
When product is less than particular pixel values, it is determined as noise region, then noise is filtered;When area is greater than particular pixel values
When, it is determined as fringe region, fringe region is reinforced.
As shown in Fig. 1, a kind of method removing picture noise provided by the invention, includes the following steps:
S01: image to be processed is obtained.
S02: the first-order difference characterisitic function g (x, y) of subchannel image calculating image.First-order difference characterisitic function g (x, y)
=‖ f (x-1, y)-f (x+1, y) ‖+‖ f (x, y-1)-f (x, y+1) ‖;Wherein, f (x, y) indicates original image, | | * | | it is norm
Or other gradient operators.
It is specifically included in this step: calculating the first-order difference characterisitic function g of image R channel image1(x,y);Calculate image B
The first-order difference characterisitic function g of channel image2(x,y);Calculate the first-order difference characterisitic function g of image G channel image3(x,y)。
S03: the corresponding binary map function B (x, y) of first-order difference characterisitic function g (x, y) in each channel image is obtained;Two
It is worth figure function B (x, y)=g (x, y) > thr;Thr indicates differential threshold.
It specifically includes: calculating the binary map function B of R channel image1(x,y);Calculate the binary map function B of channel B image2
(x,y);Calculate the binary map function B of G channel image3(x,y)。
S04: region binary map function B (x, y) in each channel image is marked using connected region labeling algorithm, the company of being formed
Logical area, and area statistics are carried out to each connected region.Connected region labeling algorithm is 8 connections or 4 connections.Each connected region is carried out
Area statistics include counting to the pixel total amount of connected region.
It specifically includes: respectively marking the corresponding region of binary map function in R channel image, channel B image and G channel image
It is denoted as m, the region for being marked as m in different channel images is marked using connected region labeling algorithm, forms x connected region, wherein
X is the integer more than or equal to 1, and m can be number or letter etc. label, as long as the effect of separator can be played.
Wherein, three channel images can be carried out while is marked in this step, it can also separately marked, setting is not
Same label symbol.Due to the corresponding binary map functional value of each channel image be it is identical, use identical label
Method can accelerate denoising process.
S05: being labeled as noise region for the connected region that area in each channel image is less than this channel corresponding region threshold value,
The connected region that area is greater than this channel corresponding region threshold value is labeled as fringe region.Wherein, this channel corresponding region threshold value refers to
Be that three channels correspond to a region threshold, the region threshold in three channels can be identical value, be also possible to not
Same value.On the other hand, region threshold is a threshold value being set in advance, the size and the degree phase of removal noise of setting
It closes, if a little bit smaller of region threshold setting can be removed big portion requiring in more stringent image noise jamming
The noise divided is specifically required according to denoising to be set.
S06: being smoothed noise region in each channel image, is sharpened processing to fringe region, will locate
Each channel image after reason synthesizes new image, the as image after removal noise.Wherein it is possible to using smothing filtering
Method is smoothed noise region.
As shown in Fig. 2, a kind of device removing picture noise provided by the invention, including first-order difference generation module,
First-order difference judgment module, connected region mark module, connected region area statistics module, discrimination module, noise processed module and side
Edge processing module;In image input first-order difference generation module to be processed, the first-order difference characterisitic function g of subchannel image is formed
(x, y), and it is transmitted to first-order difference judgment module, first-order difference judgment module generates the two of first-order difference characterisitic function g (x, y)
It is worth figure function B (x, y), and is transmitted to connected region mark module, connected region mark module is carried out according to binary map function B (x, y)
Label, forms the connected region of each channel image, and be transmitted to connected region area statistics module, connected region area statistics module system
The area of each connected region is counted, and statistical result is transmitted to discrimination module, discrimination module is according to the corresponding region in each channel
Threshold value differentiates that each connected region is noise region or fringe region, and the noise region judged is transmitted to noise processed
Module is handled, and the fringe region judged is transmitted to edge processing module and is handled, will be each after processing
Channel image composograph, the as image after removal noise.
The present invention is counted using connected region, can effectively judge sudden change region for noise region or fringe region, Ke Yi
Effective protection image edge area when handling noise can further not exist together to noise region and fringe region
Reason, while handling the amendment of noise region and fringe region well;It, can be efficiently fast using the method for subchannel image procossing noise
The noise for going channel image unless each fastly, further increases overall picture quality.
The above description is only a preferred embodiment of the present invention, and the embodiment is not intended to limit patent protection of the invention
Range, thus it is all with the variation of equivalent structure made by specification and accompanying drawing content of the invention, it similarly should be included in this
In the protection scope of invention appended claims.
Claims (9)
1. a kind of method for removing picture noise, which comprises the steps of:
S01: obtaining image to be processed, and described image is decomposed on RGB color domain, forms R channel image, the channel G figure
Picture and channel B image;
S02: the first-order difference characterisitic function g (x, y) of different channel images is calculated separately;
S03: the corresponding binary map function B (x, y) of first-order difference characterisitic function g (x, y) described in each channel image is obtained;Institute
State binary map function B (x, y)=g (x, y) > thr;The thr indicates differential threshold;
S04: marking the corresponding region binary map function B (x, y) in each channel image using connected region labeling algorithm, is formed
Connected region, and area statistics are carried out to each connected region;
S05: the connected region that area in each channel image is less than this channel corresponding region threshold value is labeled as noise region, by face
The connected region that product is greater than this channel corresponding region threshold value is labeled as fringe region;
S06: being smoothed noise region described in each channel image, is sharpened processing to the fringe region,
Each channel image after processing is synthesized into new image, the as image after removal noise.
2. a kind of method for removing picture noise according to claim 1, which is characterized in that the first-order difference characteristic letter
Number g (x, y)=‖ f (x-1, y)-f (x+1, y) ‖+‖ f (x, y-1)-f (x, y+1) ‖;Wherein, f (x, y) indicates original image, | | *
| | it is norm or gradient operator.
3. a kind of method for removing picture noise according to claim 1, which is characterized in that the connected region labeling algorithm
For 8 connections or 4 connections.
4. it is according to claim 1 it is a kind of remove picture noise method, which is characterized in that it is described to each connected region into
Row area statistics include counting to the pixel total amount of connected region.
5. a kind of method for removing picture noise according to claim 1, which is characterized in that the step S02 includes: meter
Calculate the first-order difference characterisitic function g of described image R channel image1(x,y);Calculate the first-order difference of described image channel B image
Characterisitic function g2(x,y);Calculate the first-order difference characterisitic function g of described image G channel image3(x,y)。
6. a kind of method for removing picture noise according to claim 5, which is characterized in that the step S03 includes: meter
Calculate the binary map function B of R channel image1(x,y);Calculate the binary map function B of channel B image2(x,y);Calculate G channel image
Binary map function B3(x,y)。
7. a kind of method for removing picture noise according to claim 1, which is characterized in that the step S04 includes: point
It is not m by the corresponding zone marker of binary map function in R channel image, channel B image and G channel image, is marked using connected region
It is marked as the region of m in note algorithm tag difference channel image, forms x connected region, wherein x is whole more than or equal to 1
Number.
8. a kind of method for removing picture noise according to claim 1, which is characterized in that use smooth filtering method pair
The noise region is smoothed.
9. a kind of device for removing picture noise, which is characterized in that judge mould including first-order difference generation module, first-order difference
Block, connected region mark module, connected region area statistics module, discrimination module, noise processed module and edge processing module;Wait locate
It manages image to input in the first-order difference generation module, forms the first-order difference characterisitic function g (x, y) of subchannel image, and pass
The first-order difference judgment module is transported to, the first-order difference judgment module generates the first-order difference characterisitic function g (x, y)
Binary map function B (x, y), and it is transmitted to the connected region mark module, the connected region mark module is according to binary map function B
(x, y) is marked, and forms the connected region of each channel image, and is transmitted to the connected region area statistics module, the company
Logical area's area statistics module counts the area of each connected region, and statistical result is transmitted to the discrimination module, the differentiation
Module differentiates that each connected region is noise region or fringe region according to the corresponding region threshold in each channel, and will judge
The noise region come is transmitted to the noise processed module and is handled, and the fringe region judged is transmitted to the edge
Processing module is handled, by each channel image composograph after processing, the as image after removal noise.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811415930.8A CN109636743A (en) | 2018-11-26 | 2018-11-26 | A kind of method and device removing picture noise |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811415930.8A CN109636743A (en) | 2018-11-26 | 2018-11-26 | A kind of method and device removing picture noise |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109636743A true CN109636743A (en) | 2019-04-16 |
Family
ID=66069476
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811415930.8A Pending CN109636743A (en) | 2018-11-26 | 2018-11-26 | A kind of method and device removing picture noise |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109636743A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101901477A (en) * | 2010-07-27 | 2010-12-01 | 中国农业大学 | Method and system for extracting field image edges of plant leaves |
US20140140629A1 (en) * | 2012-11-21 | 2014-05-22 | National Cheng Kung University | Methods for processing target pattern, method for generating classification system of target patterns and method for classifying detected target patterns |
CN108109123A (en) * | 2017-12-21 | 2018-06-01 | 成都微光集电科技有限公司 | A kind of image de-noising method |
-
2018
- 2018-11-26 CN CN201811415930.8A patent/CN109636743A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101901477A (en) * | 2010-07-27 | 2010-12-01 | 中国农业大学 | Method and system for extracting field image edges of plant leaves |
US20140140629A1 (en) * | 2012-11-21 | 2014-05-22 | National Cheng Kung University | Methods for processing target pattern, method for generating classification system of target patterns and method for classifying detected target patterns |
CN108109123A (en) * | 2017-12-21 | 2018-06-01 | 成都微光集电科技有限公司 | A kind of image de-noising method |
Non-Patent Citations (1)
Title |
---|
月半RAI: "几种边缘检测微分算子的比较", 《CSDN》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106651872B (en) | Pavement crack identification method and system based on Prewitt operator | |
CN106407919B (en) | Text separation method and device and image forming apparatus based on image procossing | |
CN108510491A (en) | Blur the filter method of skeleton critical point detection result under background | |
CN107729812B (en) | Method suitable for recognizing vehicle color in monitoring scene | |
Maglad | A vehicle license plate detection and recognition system | |
JP2008148298A (en) | Method and apparatus for identifying regions of different content in image, and computer readable medium for embodying computer program for identifying regions of different content in image | |
CN102956029B (en) | Image processing apparatus and image processing method | |
CN114549497A (en) | Method for detecting surface defects of walking board based on image recognition and artificial intelligence system | |
CN109345524A (en) | A kind of bearing open defect detection system of view-based access control model | |
CN102930537A (en) | Image detection method and system | |
WO2012111906A1 (en) | Apparatus and method for generating image feature data | |
Zhang et al. | A combined algorithm for video text extraction | |
CN105336035B (en) | A kind of method and system of dirty crown word number image classification | |
CN107798304B (en) | Method for rapidly auditing video | |
CN110335233A (en) | Express-way guard-rail plates defect detecting system and method based on image processing techniques | |
CN102567969B (en) | Color image edge detection method | |
CN111080723A (en) | Image element segmentation method based on Unet network | |
CN102768758B (en) | Improved color image unsharp masking (USM) method | |
CN113012156B (en) | Intelligent solid wood board color classification method | |
CN106778777A (en) | A kind of vehicle match method and system | |
CN109636743A (en) | A kind of method and device removing picture noise | |
CN102254173B (en) | Method for segmenting fingerprint image based on combination of k-means clustering and linear neighborhood propagation | |
CN104103039B (en) | Image noise estimation method | |
CN108805838A (en) | A kind of image processing method, mobile terminal and computer readable storage medium | |
CN102135503B (en) | Method for detecting color overprinting deviation by using image edge information |
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 |