CN102118547A - Image weighted filtering method - Google Patents
Image weighted filtering method Download PDFInfo
- Publication number
- CN102118547A CN102118547A CN 201110076651 CN201110076651A CN102118547A CN 102118547 A CN102118547 A CN 102118547A CN 201110076651 CN201110076651 CN 201110076651 CN 201110076651 A CN201110076651 A CN 201110076651A CN 102118547 A CN102118547 A CN 102118547A
- Authority
- CN
- China
- Prior art keywords
- filtering
- image
- pixel
- luminance component
- template
- 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
Images
Landscapes
- Picture Signal Circuits (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an image weighted filtering method relating to the processing of digital video images. The image weighted filtering method includes the following steps: setting the threshold value and the weighting coefficient of a noise point; separating a luminance component out of an image signal; judging the property of a pixel point to be processed; and processing the pixel point by an undirected filtering module and/or a directed filtering module according to the property of the pixel point. The image weighting-filtering method can maintain clear edges and details of an image simultaneously when eliminating image noise, thereby greatly improving the picture quality effect of image processing. The image weighted filtering method realizes the processing of both video images and static images, thus having a very broad applicable range.
Description
Technical field
The present invention relates to the processing of digital video image, a kind of specifically method of image weighted filtering.
Background technology
Image filtering is that the key step of eliminating noise in the field of video image processing gathers.Image filtering is that the noise to target image suppresses under the condition that as far as possible keeps the image detail feature, is indispensable operation in the image preliminary treatment, and the quality of its treatment effect will directly be rung validity and the reliability of handling and analyzing to successive image.In image processing, the attribute of pixel comprises: noise, image and background.Noise often shows as one and causes the isolated pixel point or the pixel block of visual effect by force on image.Generally, noise signal is uncorrelated with the object that will study, and it occurs with useless message form, upsets the observable information of image.For data image signal, psophometer is either large or small extreme value, these extreme values act on the true gray value of image pixel by plus-minus, cause bright, dim spot interference at image, greatly reduced picture quality, influenced image restoration, cut apart, the carrying out of follow-up work such as feature extraction, figure identification.For this reason, people have successively made up multiple filtering and noise reduction method from spatial domain and frequency domain or the overall situation and part, typical method has Wiener filtering, least square method filtering, the multiple image method of average, medium filtering and weighted median filtering etc., the attribute of judging pixel normally preestablishes a threshold value according to l-G simulation test, ask average back and the threshold value that sets to compare to the value of the pixel around this pixel then, if mean value>threshold value, judge that this pixel is a noise, if mean value≤threshold value judges that this pixel is background or image.But the method for this judgement pixel attribute makes that easily result's error is bigger, and simultaneously at present numerous noise-removed filtering methods also all exists a defective, when obtaining preferably the smothing filtering effect, is difficult to keep simultaneously image border and details that is:.So, explore and a kind ofly can satisfy realtime graphic and handle and to reach the filtering and noise reduction purpose, can keep the new method of image border and details again, be an important directions of image filtering denoising development.
Summary of the invention
The invention provides a kind of method of image filtering, this method can also keep image border and details clear in the removal of images noise after by the mode of weighting image being handled.
The method of image weighted filtering of the present invention comprises: threshold value and the weight coefficient of setting noise spot; Isolate the luminance component in the picture signal; Judge the attribute of pending pixel; Use undirected filtering template or use undirected filtering template simultaneously and oriented filtering template is handled this pixel according to the attribute of pending pixel.Threshold value and weight coefficient are the parameters of presetting, and its value can be made as and realize that effect is worth preferably.These values can be passed through I in actual applications
2The C communication interface is made amendment.The undirected filtering template of using in pixel is handled is to be center pixel with pending pixel, and pixel on every side has identical influence to it, and oriented filtering template is one or more templates with certain angle.In the method, investigate the weight difference of different angles pixel and central pixel point, and then judge that central pixel point is a noise spot by calculating.When if the attribute of pending pixel is noise, use undirected filtering template to handle, if when the attribute of pending pixel is image or background, use undirected filtering template and oriented filtering template to handle.Through such processing, when removing noise spot, can also keep image detail.
In order to make image filtering reach better effect, the undirected filtering template of use is 1, and oriented filtering template is 8, and wherein the angle of 8 oriented templates is respectively 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °.The filtering template is generally square shape, on the filtering template, be the center with pending pixel, the forward of X-axis is 0 °, the straight line that then connects center pixel (pending pixel) and each neighbor only may be four of the AABBCCDD shown in a of Fig. 1, also represented all possible direction, 8 promptly above-mentioned angles.By all angles being carried out omnidirectional's weighted pixel, can obtain better effect.
In the process of processed pixels point, a kind of mode of preferably handling according to the attribute of pending pixel comprises step:
A. calculate the absolute value of the difference of the luminance component average of other pixel in the luminance component of pending pixel and undirected filtering template and each oriented filtering template;
B. obtain the minimal difference among the step a;
If c. the threshold value of the noise spot of the minimal difference>described setting among the step b is used undirected filtering template to calculate new luminance component, and the luminance component of pending pixel is replaced with new luminance component; If the threshold value of the noise spot of the minimal difference among the step b≤described setting is used undirected filtering template and each oriented filtering template to calculate corresponding weighted mean, and the luminance component of pending pixel is replaced with described weighted mean.
In image processing, the attribute of pixel comprises: noise, image and background.If the threshold value of the noise spot of the minimal difference among the step b>described setting, the determined property of pending pixel are noise; If the threshold value of the noise spot of the minimal difference among the step b≤described setting, the determined property of pending pixel are background or image.Method of the present invention has carried out judging more accurately to the pixel of different attribute just, makes the accuracy of judging be significantly improved, and the pixel of different attribute is handled targetedly, reaches the purpose of filtering.
A kind of concrete scheme is that pending pixel by the conversion of video color difference signal chrominance space, is isolated the luminance component in the picture signal.With the value that the pixel of different attribute calculates through separately method, replacement pixel is put original luminance component, finishes the processing to pixel.
Further scheme is that the number and the template number of weight coefficient adapt.Weight coefficient is designated k
Ab, wherein a is 〉=0 integer, the quantity of span and filtering template adapts, if the ading up to of filtering template 9 (1 undirected filtering template and 8 oriented filtering templates), the value of a is 0~8 integer.The value of b is 0~4 integer, so k
AbValue be embodied in k
00, k
01, k
02..., k
10, k
11, k
12... by that analogy.
Verify that on l-G simulation test and image algorithm development board the method for image weighted filtering of the present invention can also keep image border and details clear, has improved the image quality effect of image processing greatly in the removal of images noise.And this method can be handled the processing and the still image of video image and handle, and applicable surface is very extensive.
Below in conjunction with embodiment, foregoing of the present invention is described in further detail again by the accompanying drawing illustrated embodiment.But this should be interpreted as that the scope of the above-mentioned theme of the present invention only limits to following example.Do not breaking away under the above-mentioned technological thought situation of the present invention, various replacements or change according to ordinary skill knowledge and customary means are made all should comprise within the scope of the invention.
Description of drawings
Fig. 1 is the undirected filtering template and the oriented filtering template schematic diagram of the method for image weighted filtering of the present invention.
Fig. 2 is the flow chart of the method for image weighted filtering of the present invention.
A among Fig. 1: undirected filtering template; B:0 ° of orientation filtering template; C:45 ° of orientation filtering template; D:90 ° of orientation filtering template; E:135 ° of orientation filtering template; F:180 ° of orientation filtering template; G:225 ° of orientation filtering template; H:270 ° of orientation filtering template; I:315 ° of orientation filtering template.
Embodiment
Expressed undirected filtering template and 8 oriented filtering templates of orientation angles among Fig. 1, on the basis of these filtering templates, carried out image filtering and handle.
As shown in Figure 2, set threshold value V and every weight coefficient k of noise spot
AbThe integer of a=0~8 wherein, the integer of b=0~4 can pass through I in actual applications
2The C communication structure is revised preset threshold V and every weight coefficient k
AbWith the video signal of input,, isolate pending pixel f (x, luminance component Y y) through the video color difference signal chrominance space conversion of YUV or YcbCr mode.
Judge that (that is: (x y) belongs to any in image, background or the noise to pixel f to image slices vegetarian refreshments f, and uses different filtering templates to carry out computing to the pixel of different attribute for x, attribute y).The method of its judgement and processing is:
1) (x removes pending pixel f (x, y) the absolute value V of the difference of the average of Wai other pixel intensity component Y value in each filtering template among the value of luminance component Y y) and Fig. 1 to calculate pending pixel f respectively
i(i=0,1,2,3,4,5,6,7,8), the i value 0~8 with Fig. 1 in a~i filtering template corresponding one by one, that is: V
0Corresponding templates a, V
1Corresponding templates b ..., and the like.V
iThe algorithm of each value is:
V
0=|[f(x,y-1)+f(x-1,y)+f(x,y+1)+f(x+1,y)]/4-f(x,y)|
V
1=|[f(x-1,y+1)+f(x,y+1)+f(x,y+2)+f(x+1,y+1)]/4-f(x,y)|
V
2=|[f(x-1,y)+f(x-1,y+1)+f(x-2,y+2)+f(x,y+1)]/4-f(x,y)|
V
3=|[f(x-1,y-1)+f(x-2,y)+f(x-1,y)+f(x-1,y+1)]/4-f(x,y)|
V
4=|[f(x-2,y-2)+f(x-1,y-1)+f(x,y-1)+f(x-1,y)]/4-f(x,y)|
V
5=|[f(x,y-2)+f(x-1,y-1)+f(x,y-1)+f(x+1,y-1)]/4-f(x,y)|
V
6=|[f(x+2,y-2)+f(x,y-1)+f(x+1,y-1)+f(x+1,y)]/4-f(x,y)|
V
7=|[f(x+1,y-1)+f(x+1,y)+f(x+1,y+1)+f(x+2,y)]/4-f(x,y)|
V
8=|[f(x+1,y)+f(x,y+1)+f(x+1,y+1)+f(x+2,y+2)]/4-f(x,y)|
2) get V
iMinimum value, that is:
V
min=min{V
0,V
1,V?
2,V?
3,V?
4,V?
5,V?
6,V?
7,V?
8}
3) if V
Min>V, that is: remove in each filtering template pending pixel f (x, y) outside, (x, the absolute value of the difference of luminance component Y value y) are greater than threshold value V, and (x y) is isolated noise point to then pending pixel f for the arithmetic mean of all pixel Y values and pending pixel f.At this moment, be calculated as follows new luminance component Y value with template a and substitute pending pixel f (x, luminance component Y value y).
V
0=int|[f (x, y-1)+f (x-1, y)+f (x, y+1)+f (x+1, y)]/4-f (x, y) | (int is for rounding algorithm).
4) if V
Min≤ V, that is: remove in each filtering template pending pixel f (x, y) outside, (x, the absolute value of the difference of luminance component Y value y) are less than threshold value V, and (x y) belongs to image or background to then pending pixel f for the arithmetic mean of all pixel Y values and pending pixel f.At this moment, be calculated as follows weighted mean L that there emerged a the sequence number corresponding templates and substitute pending pixel f (x, luminance component Y value y).
L
0=[k
00f(x,y-1)+k
01f(x-1,y)+k
02f(x,y+1)+k
03f(x+1,y)+k
04f(x,y)]/(k
00+k
01+k
02+k
03+k
04)
L
1=[k
10f(x-1,y+1)+k
11f(x,y+1)+k
12f(x,y+2)+k
13f(x+1,y+1)+k
14f(x,y)]/(k
10+k
11+k
12+k
13+k
14)
L
2=[k
20f(x-1,y)+k
21f(x-1,y+1)+k
22f(x-2,y+2)+k
23f(x,y+1)+k
24f(x,y)/(k
20+k
21+k
22+k
23+k
24)
L
3=[k
30f(x-1,y-1)+k
31f(x-2,y)+k
32f(x-1,y)+k
33f(x-1,y+1)+k
34f(x,y)]/(k
30+k
31+k
32+k
33+k
34)
L
4=[k
40f(x-2,y-2)+k
41f(x-1,y-1)+k
42f(x,y-1)+k
43f(x-1,y)+k
44f(x,y)]/(k
40+k
41+k
42+k
43+k
44)
L
5=[k
50f(x,y-2)+k
51f(x-1,y-1)+k
52f(x,y-1)+k
53f(x+1,y-1)+k
54f(x,y)]/(k
50+k
51+k
52+k
53+k
54)
L
6=[k
60f(x+2,y-2)+k
61f(x,y-1)+k
62f(x+1,y-1)+k
63f(x+1,y)+k
64f(x,y)]/(k
60+k
61+k
62+k
63+k
64)
L
7=[k
70f(x+1,y-1)+k
70f(x+1,y)+k
72f(x+1,y+1)+k
73f(x+2,y)+k
74f(x,y)]/(k
70+k
71+k
72+k
73+k
74)
L
8=[k
80f(x+1,y)+k
81f(x,y+1)+k
82f(x+1,y+1)+k
83f(x+2,y+2)+k
84f(x,y)]/(k
80+k
81+k
82+k
83+k
84)
Above k in various
Ab(a=0,1,2,3,4,5,6,7,8; B=0,1,2,3,4) be weight coefficient, set k according to the actual experiment effect of image algorithm emulation and development board
AbValue, for example: (x, y) inverse of the distance between is as the weight coefficient of this pixel can to adopt in the filtering template each pixel and pending pixel f.
The new value that calculates is replaced pending pixel f, and (x behind luminance component Y y), repeats above step with the traversal entire image, finishes the filtering and noise reduction of all pixels and handles.At last the image after the Filtering Processing denoising is exported.
Claims (6)
1. the method for image weighted filtering is characterized by and comprises: threshold value and the weight coefficient of setting noise spot; Isolate the luminance component in the picture signal; Judge the attribute of pending pixel; Use undirected filtering template or use undirected filtering template simultaneously and oriented filtering template is handled this pixel according to the attribute of pending pixel.
2. the method for image weighted filtering as claimed in claim 1, it is characterized by described undirected filtering template is 1, oriented filtering template is 8, and wherein the angle of 8 oriented filtering templates is respectively 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °.
3. the method for image weighted filtering as claimed in claim 1 or 2 is characterized by to handle according to the attribute of pending pixel and comprises step:
A. calculate the absolute value of the difference of the luminance component average of other pixel in the luminance component of pending pixel and undirected filtering template and each oriented filtering template;
B. obtain the minimal difference among the step a;
If c. the threshold value of the noise spot of the minimal difference>described setting among the step b is used undirected filtering template to calculate new luminance component, and the luminance component of pending pixel is replaced with new luminance component; If the threshold value of the noise spot of the minimal difference among the step b≤described setting is used undirected filtering template and each oriented filtering template to calculate corresponding weighted mean, and the luminance component of pending pixel is replaced with described weighted mean.
4. the method for image weighted filtering as claimed in claim 1, the attribute that it is characterized by described pending pixel comprises: noise, image and background.
5. the method for image weighted filtering as claimed in claim 1 is characterized by by the conversion of video color difference signal chrominance space, isolates the luminance component in the described picture signal.
6. the method for image weighted filtering as claimed in claim 1, the number and the template number that it is characterized by described weight coefficient adapt.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110076651 CN102118547A (en) | 2011-03-29 | 2011-03-29 | Image weighted filtering method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110076651 CN102118547A (en) | 2011-03-29 | 2011-03-29 | Image weighted filtering method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN102118547A true CN102118547A (en) | 2011-07-06 |
Family
ID=44217129
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201110076651 Pending CN102118547A (en) | 2011-03-29 | 2011-03-29 | Image weighted filtering method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102118547A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103366344A (en) * | 2012-03-27 | 2013-10-23 | 富士通株式会社 | Method and device for edge-preserving filtering |
CN105894464A (en) * | 2016-03-28 | 2016-08-24 | 福州瑞芯微电子股份有限公司 | Median filtering image processing method and apparatus |
CN107563981A (en) * | 2017-09-04 | 2018-01-09 | 哈尔滨理工大学 | A kind of evaluation method of image ringing effect |
CN108269237A (en) * | 2016-12-30 | 2018-07-10 | 华为技术有限公司 | A kind of image filtering device, system and method |
CN110517217A (en) * | 2019-01-23 | 2019-11-29 | 任成付 | Computer overall data identifying platform |
CN111666854A (en) * | 2020-05-29 | 2020-09-15 | 武汉大学 | High-resolution SAR image vehicle target detection method fusing statistical significance |
CN113129220A (en) * | 2019-12-31 | 2021-07-16 | 荣耀终端有限公司 | Image processing method and electronic device |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1633159A (en) * | 2005-01-26 | 2005-06-29 | 北京中星微电子有限公司 | A method for removing image noise |
WO2007074964A1 (en) * | 2005-12-29 | 2007-07-05 | Mtekvision Co., Ltd | Device for removing noise in image data |
CN101237523A (en) * | 2008-02-25 | 2008-08-06 | 宝利微电子***控股公司 | Main edge detection method and noise reduction method and device based on this method |
-
2011
- 2011-03-29 CN CN 201110076651 patent/CN102118547A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1633159A (en) * | 2005-01-26 | 2005-06-29 | 北京中星微电子有限公司 | A method for removing image noise |
WO2007074964A1 (en) * | 2005-12-29 | 2007-07-05 | Mtekvision Co., Ltd | Device for removing noise in image data |
CN101237523A (en) * | 2008-02-25 | 2008-08-06 | 宝利微电子***控股公司 | Main edge detection method and noise reduction method and device based on this method |
Non-Patent Citations (2)
Title |
---|
《东北大学学报》 20070731 张刚 等 基于内容图像检索中针对显著方向的有向滤波器 第978-981页 1-6 第28卷, 第7期 * |
《计算机工程与设计》 20100131 杨吉宏 等 灰度图像加权有向平滑滤波算法 第153-156页 1-6 , 第1期 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103366344A (en) * | 2012-03-27 | 2013-10-23 | 富士通株式会社 | Method and device for edge-preserving filtering |
CN105894464A (en) * | 2016-03-28 | 2016-08-24 | 福州瑞芯微电子股份有限公司 | Median filtering image processing method and apparatus |
CN105894464B (en) * | 2016-03-28 | 2018-08-31 | 福州瑞芯微电子股份有限公司 | A kind of medium filtering image processing method and device |
CN108269237A (en) * | 2016-12-30 | 2018-07-10 | 华为技术有限公司 | A kind of image filtering device, system and method |
CN108269237B (en) * | 2016-12-30 | 2021-11-19 | 华为技术有限公司 | Image filtering device, system and method |
CN107563981B (en) * | 2017-09-04 | 2020-08-21 | 哈尔滨理工大学 | Method for evaluating image ringing effect |
CN107563981A (en) * | 2017-09-04 | 2018-01-09 | 哈尔滨理工大学 | A kind of evaluation method of image ringing effect |
CN110517217B (en) * | 2019-01-23 | 2020-04-28 | 莱芜职业技术学院 | Computer integral data recognition device |
CN110517217A (en) * | 2019-01-23 | 2019-11-29 | 任成付 | Computer overall data identifying platform |
CN113129220A (en) * | 2019-12-31 | 2021-07-16 | 荣耀终端有限公司 | Image processing method and electronic device |
CN113129220B (en) * | 2019-12-31 | 2024-03-29 | 荣耀终端有限公司 | Image processing method and electronic equipment |
CN111666854A (en) * | 2020-05-29 | 2020-09-15 | 武汉大学 | High-resolution SAR image vehicle target detection method fusing statistical significance |
CN111666854B (en) * | 2020-05-29 | 2022-08-30 | 武汉大学 | High-resolution SAR image vehicle target detection method fusing statistical significance |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109870461B (en) | Electronic components quality detection system | |
US10521885B2 (en) | Image processing device and image processing method | |
CN102118547A (en) | Image weighted filtering method | |
CN107742291B (en) | Defect detection method and device for photovoltaic glass | |
CN106023204B (en) | A kind of method and system removing mosquito noise based on edge detection algorithm | |
CN107301624B (en) | Convolutional neural network defogging method based on region division and dense fog pretreatment | |
CN107240084B (en) | Method and device for removing rain from single image | |
WO2015190136A1 (en) | Image processing system and computer-readable recording medium | |
CN104504669B (en) | A kind of medium filtering detection method based on local binary patterns | |
CN111008961B (en) | Transmission line equipment defect detection method and system, equipment and medium thereof | |
CN109377450B (en) | Edge protection denoising method | |
CN109060842B (en) | Citrus surface defect detection method based on surface fitting correction | |
CN111429370A (en) | Method and system for enhancing images in coal mine and computer storage medium | |
CN112528868B (en) | Illegal line pressing judgment method based on improved Canny edge detection algorithm | |
CN112862753B (en) | Noise intensity estimation method and device and electronic equipment | |
Yu et al. | Image and video dehazing using view-based cluster segmentation | |
CN103942756B (en) | A kind of method of depth map post processing and filtering | |
CN105894474A (en) | Non-linear image enhancement method, and edge detection method using the same | |
CN111179186A (en) | Image denoising system for protecting image details | |
CN100367770C (en) | Method for removing isolated noise point in video | |
CN114332081B (en) | Textile surface abnormity determination method based on image processing | |
CN106651816B (en) | Method and system for halftone processing of digital image | |
Chen et al. | Improve transmission by designing filters for image dehazing | |
CN104103037A (en) | Image enhancement processing method and device | |
CN106846262B (en) | Method and system for removing mosquito noise |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20110706 |