CN101854468B - Repairing method for removing image noise - Google Patents
Repairing method for removing image noise Download PDFInfo
- Publication number
- CN101854468B CN101854468B CN2009101336165A CN200910133616A CN101854468B CN 101854468 B CN101854468 B CN 101854468B CN 2009101336165 A CN2009101336165 A CN 2009101336165A CN 200910133616 A CN200910133616 A CN 200910133616A CN 101854468 B CN101854468 B CN 101854468B
- Authority
- CN
- China
- Prior art keywords
- pixel
- color
- green
- revised
- red
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Color Television Image Signal Generators (AREA)
Abstract
The present invention discloses a repairing method for removing image noise, which is used for repairing noise generated by Bayer pattern color interpolation treatment. The repairing method comprises the following steps: selecting parts of regions in a Bayer pattern to set a selected region; selecting a plurality of color units from the selected region one by one, wherein any one color unit at least comprises a green pixel, a red pixel and a blue pixel; orderly adjusting the color weight of the green pixel, the red pixel and the blue pixel; and according to each color weight, compensating pixels in the selected region to produce a repaired pixel.
Description
Technical field
The present invention relates to a kind of image method for repairing and mending, be particularly to a kind of image noise method for repairing and mending through the digitized video of Bel segment noise that color treatments produces.
Background technology
In photosensory assembly, the array of being made up of millions of photosensitive units (or pixel) covers the surface of transducer.Stored charge after being subjected to light on the array of pixels of whole photosensory assembly is read at an end of photosensory assembly.(Analog Front End, AFE) chip or sensitization processor quantize by an AFE (analog front end) with stored charge.
In order accurately to show a width of cloth coloured image, all need 3 kinds of color cards on each location of pixels of transducer, it is the most common that what be used is exactly the RGB three primary colors.But, can make the cost of digital camera significantly increase if three layers of color photosensory assembly are set on same location of pixels.Carry out pixel color and be subjected to light so someone proposes the mode of colored filter array (color filter array is hereinafter to be referred as CFA).The most widely used CFA is Bel's segment (Bayer pattern) now.
Please refer to shown in Figure 1ly, it is Bel's segment schematic diagram.Bel's segment 100 has utilized human eye that green intensity distinguished that sensitivity is greater than red or blue principle.Therefore, in the colored filter array of Bel's segment 100, the number of times that green color filter occurs is the twice of blue color filter or Red lightscreening plate, makes per four pixels form a unit.Putting in order of filter is Red lightscreening plate and alternately arrangement of green color filter; Inferior row then are alternately to be arranged by green color filter and blue color filter.
The light quantity that is received according to each pixel by the digital camera processor is carried out the interpolation processing of color at last.Please refer to shown in the table 1, it is the part tabulation of the optical filtering array of pixel:
G 1 | R 2 | G 3 | R 4 |
B 5 | G 6 | B 7 | G 8 |
G 9 | R 10 | G 11 | R 12 |
B 13 | G 14 | B 15 | G 16 |
Table 1
Be calculated as example with interpolation, for the green pixel interpolation of blue pixel and red pixel position.Please refer to shown in the table 1, because four pixels of tie around all have real green pixel, therefore to reduce this green pixel that loses, green pixel arround available carries out interpolation and repairs, in like manner if red pixel and blue pixel are when losing, and the identical color in also available its adjacent domain is done interpolation and repaired:
G′
7=(G
3+G
6+G
8+G
11)/4,
R′
7=(R
2+R
4+R
10+R
12)/4,
B′
6=(B
5+B
7)/2,
G wherein, R, B represent real green pixel, red pixel and blue pixel respectively; Green pixel values, red pixel value and blue pixel value that on behalf of interpolation, G ', R ' and B ' then obtain.
After above-mentioned color of pixel is rebuild, can obtain Bel's segment 100 all color of pixel correction values.Obtaining following table 2 after overcompensation, it is each colors list of each pixel after compensating.
G 1R′ 1B′ 1 | G′ 2R 2B′ 2 | G 3R′ 3B′ 3 | G′ 4R 4B′ 4 |
G′ 5R′ 5B 5 | G 6R′ 6B′ 6 | G′ 7R′ 7B 7 | G 8R′ 8B′ 8 |
G 9R′ 9B′ 9 | G′ 10R 10B′ 10 | G 11R′ 11B′ 11 | G′ 12R 12B′ 12 |
G′ 13R′ 13B 13 | G 14R′ 14B′ 14 | G′ 15R′ 15B 15 | G 16R′ 16B′ 16 |
Table 2
But only utilize single kind of color to come, can occur longer-distance colored pixels easily and can have a strong impact on this colored pixels compensating and correcting.In other words, because be not that the colored pixels of all Bel's segments 100 is all similar in the digitized video.So the neighbor that otherness is bigger will have influence on whole average.At this moment will produce the noise in the digitized video.
If to each compensation pixel carry out denoising and handle, can be because this pixel have been passed through the compensation of other color.So the effect to denoising has its restriction.
Summary of the invention
In view of above problem, main purpose of the present invention is to provide a kind of method for repairing and mending of removing image noise, is used to repair Bel's segment (Bayer pattern) color interpolation and handles the noise that is produced.
For reaching above-mentioned purpose, the present invention discloses a kind of method for repairing and mending of removing image noise, and method for repairing and mending comprises the following steps: to choose subregion in Bel's segment to set a chosen area; From the chosen area unit that gets colors one by one, comprise green pixel, red pixel and blue pixel in the color cell at least; Green pixel is carried out the first color weight adjustment, revised green filtering parameter in order to produce one; Utilization has been revised green filtering parameter red pixel has been carried out the second color weight adjustment, has revised red filtering parameter in order to produce; Utilize and to have revised green filtering parameter, revise red filtering parameter blue pixel is carried out the 3rd color weight adjustment, revised the blue color filtered parameter in order to produce; According to revising green filtering parameter, revise red filtering parameter and revising the blue color filtered parameter is carried out color to the pixel of relevant position compensation.
Wherein according to following formula to green pixel, carry out the first color weight adjustment, revised green filtering parameter:
G
recovery=Filter
g(G);
Red pixel is according to following formula and revise green filtering parameter, carries out the second color weight adjustment, has been revised red filtering parameter:
R
recovery=Filter
r(R,G
recovery);
Blue pixel according to following formula, revise green filtering parameter with revise red filtering parameter, carry out the 3rd color weight adjustment, revised the blue color filtered parameter:
B
recovery=Filter
r(B,R
recovery,G
recovery)。
The invention provides a kind ofly, carry out reinforcement colored pixels to the raw video data that image sensor received.For each colored pixels, the present invention all can calculate its interpolation of carrying out the adjacent color pixel, makes this colored pixels not only only be referenced to the same color pixel.Use the noise that reduces in the digitized video that is caused because of the existing colour-difference added time.
Relevant characteristics and implementation of the present invention cooperate diagram to be described in detail as follows as most preferred embodiment now.
Description of drawings
Fig. 1 is Bel's segment schematic diagram of prior art;
Fig. 2 is an operation workflow schematic diagram of the present invention;
Fig. 3 adjusts schematic diagram for the present invention to color weight;
The schematic diagram of Fig. 4 a for green pixel is compensated;
The schematic diagram of Fig. 4 b for red pixel is compensated;
The schematic diagram of Fig. 4 c for blue pixel is compensated;
Fig. 4 d is the schematic diagram to each colored pixels array.
[primary clustering symbol description]
100 Bel's segments
300 Bel's segments
311 green pixels
312 red pixels
313 blue pixel
420 green level and smooth shieldings
421 green pixel arrays
Green pixel array after 422 compensation
430 red level and smooth shieldings
431 red pixel arrays
Red pixel array after 432 compensation
440 blue level and smooth shieldings
441 blue pixel arrays
Blue pixel array after 442 compensation
Embodiment
Please refer to shown in Figure 2ly, it is an operation workflow schematic diagram of the present invention.Method for repairing and mending of the present invention comprises the following steps: to set chosen area (step S210), in order to choose the subregion in Bel's segment.Chosen area is a m*n array of pixels, and chosen area comprises a color cell at least.Each color cell has comprised green pixel, red pixel and blue pixel.
Choose a plurality of color cell (step S220) from chosen area, comprise green pixel, red pixel and blue pixel in each color cell.Calculate the signal to noise ratio (Signal-to-Noise Rate) of respective color respectively according to each colored pixels in Bel's segment 300, it is respectively green pixel signal to noise ratio, red pixel signal to noise ratio and blue pixel signal to noise ratio.Signal to noise ratio refers to the ratio of available signal power (Power of Signal) and-noise power (Power ofNoise).Please refer to formula 1, it is the computing formula of signal to noise ratio:
P
SignalBe signal power (Power of Signal); P
NoiseBe noise power (Power ofNoise); The digitized video size is the M*N array of pixels; X
IjBe (i, j) the undisturbed ideal signal in position in the image; Z
IjThe signal that obtains by the position of reality.Generally speaking, the signal of the high more representative of snr value is clear more.For digitized video, the high more then quality of image of signal to noise ratio is clear more.
Green pixel is carried out the first color weight adjustment (step S230), revised green filtering parameter in order to produce.Carry out the color weight adjustment according to 2 pairs of green pixels of following formula:
G
Recovery=Filter
g(G) formula 2
G
Recovery: for revising green filtering parameter; Filter
g: be the green channel filter.
Then, utilization has been revised green filtering parameter red pixel has been carried out the second color weight adjustment, has revised red filtering parameter (step S240) in order to produce.Red pixel carries out the second color weight adjustment according to following formula 3 and the green filtering parameter of the correction of formula 2 gained:
R
Recovery=Filter
r(R, G
Recovery) formula 3
R
Recovery: for revising red filtering parameter; Filter
r: be the red channel filter.
Utilize and to have revised green filtering parameter, revise red filtering parameter blue pixel is carried out the 3rd color weight adjustment, revised blue color filtered parameter (step S250) in order to produce.Blue pixel according to following formula 4, revise green filtering parameter with revise red filtering parameter, carry out the 3rd color weight adjustment:
B
Recovery=Filter
b(B, R
Recovery, G
Recovery) formula 4
B
Recovery: for revising the blue color filtered parameter; Filter
b: be the blue channel filter.
At last, according to revising green filtering parameter G
Recovery, revise red filtering parameter R
RecoveryWith revise the blue color filtered B parameter
RecoveryThe pixel of relevant position is carried out the compensation (step S260) of color.And please cooperate with reference to shown in Figure 3, it adjusts schematic diagram for the present invention to color weight.Each Color Channel filter can see through the mode of spatial domain and (for example: shielding etc.) directly revise pixel value; Or the mode that sees through frequency domain comes correspondent frequency in the filtering digitized video (for example: Fast Fourier transform, be called for short FFT, discrete cosine transform, be called for short DCT or Discrete Wavelet Transformation, be called for short DWT etc.).Because green pixel 311 shared data quantity in Bel's segment 300 is the twice of red pixel 312 and blue pixel 313.So the signal to noise ratio of green pixel 311 is high compared to the signal to noise ratio of red pixel 312 or blue pixel 313.Therefore can utilize green pixel 311 to adjust the weighted value of red pixel 312 and blue pixel 313.
Please refer to shown in Fig. 4 a, Fig. 4 b and Fig. 4 c schematic diagram that it compensates for each colored pixels of the present invention.Explain orally the present invention at this for clear, with level and smooth shielding as embodiment.
Utilize three noises in the level and smooth shielding removal image respectively.Each shielding is the 3*3 array size, and array element is in order to adjust the pixel value of relevant position.Three level and smooth shieldings are respectively: green level and smooth shielding 420, red level and smooth shielding 430 and blue level and smooth shielding 440, wherein green level and smooth shielding 420, red level and smooth shielding 430 are 1 with the summation of blue level and smooth shielding 440.
The plane coordinates of left is represented a green pixel array 421 among Fig. 4 a, and utilizes the green pixel array 421 of chosen area circle choosing part.Then, the result according to following formula 2 shields 420 in order to produce the green level and smooth of respective weights.The green level and smooth shielding 420 of original green pixel array 421 processes is smoothly calculated.At last, produce green pixel array 422 after the compensation right-hand among Fig. 4 a.
Then produce red level and smooth shielding 430 according to the signal to noise ratio of the green pixel array 422 after the compensation, green pixel 311 and the signal to noise ratio of red pixel 312.The plane coordinates of left is represented a red pixel array 431 in Fig. 4 b, and it is similar to green pixel 311 that it smoothly shields the smoothing processing process, so do not add to give unnecessary details.Producing blue smoothly shielding 440 according to the signal to noise ratio of the green pixel array 422 after the compensation, green pixel 311 and the signal to noise ratio of red pixel array 432 after the compensation and red pixel 312.Utilizing 440 pairs of blue pixel 313 of blue level and smooth shielding to carry out color compensating again handles.The plane coordinates of left is represented a blue pixel array 441 among Fig. 4 c, and according to the level and smooth shielding 440 of above-mentioned resulting blueness, and produce blue pixel array 442 after compensating with reference to the signal to noise ratio of the signal to noise ratio of the signal to noise ratio of green pixel 311 and red pixel 312, blue pixel 313.Please refer to shown in Fig. 4 d, it is the schematic diagram to each colored pixels array.And utilize corresponding level and smooth shielding to carry out its color compensating respectively each colored pixels array, use the generation that reduces noise pixel.
The invention provides a kind ofly, carry out reinforcement colored pixels to the raw video data that image sensor received.For each colored pixels, the present invention all can calculate its interpolation of carrying out the adjacent color pixel, makes this colored pixels not only only be referenced to the same color pixel.Further utilize other colored pixels to carry out the weight adjustment of current color, use the noise that reduces in the digitized video that Cheng Shisuo causes because colour-difference is made amends for one's faults.
Though the present invention discloses as above with a preferred embodiment; right its is not in order to limit the present invention; under the situation that does not deviate from spirit of the present invention and essence thereof; those of ordinary skill in the art work as can make various corresponding changes and distortion according to the present invention, but these corresponding changes and distortion all should belong to the protection range of the appended claim of the present invention.
Claims (4)
1. a method for repairing and mending of removing image noise is used to repair Bel's segment (Bayer pattern) color interpolation and handles the noise that is produced, and it is characterized in that this method for repairing and mending comprises the following steps:
Choose subregion in this Bel's segment to set a chosen area;
Choose a plurality of color cell one by one from this chosen area, comprise a green pixel, a red pixel and a blue pixel in each this color cell at least;
This green pixel is carried out one first color weight adjustment, revised green filtering parameter in order to produce one;
Utilize this to revise green filtering parameter this red pixel is carried out one second color weight adjustment, revised red filtering parameter in order to produce one;
Utilize this to revise green filtering parameter, this has been revised red filtering parameter this blue pixel has been carried out one the 3rd color weight adjustment, has revised the blue color filtered parameter in order to produce one; And
According to this revised green filtering parameter, this has revised red filtering parameter and this has revised the blue color filtered parameter is carried out color to the pixel of relevant position compensation.
2. the method for repairing and mending of removal image noise as claimed in claim 1 is characterized in that, this chosen area is a m*n array of pixels, and this chosen area comprises a color cell at least.
3. the method for repairing and mending of removal image noise as claimed in claim 1, it is characterized in that, after choosing the step of these color cell one by one, also comprise the following steps: to obtain a green pixel signal to noise ratio, a red pixel signal to noise ratio and a blue pixel signal to noise ratio in this Bel's segment from this chosen area.
4. the method for repairing and mending of removal image noise as claimed in claim 1 is characterized in that, according to following formula this green pixel is carried out this first color weight adjustment, has revised green filtering parameter to produce this:
G
recovery=Filter
g(G),
Revised green filtering parameter according to following formula and this, this red pixel carried out this second color weight adjustment, revised red filtering parameter to produce this:
R
recovery=Filter
r(R,G
recovery),
According to following formula, this has revised green filtering parameter and this has revised red filtering parameter, and this blue pixel is carried out the 3rd color weight adjustment, has revised the blue color filtered parameter to produce this:
B
recovery=Filter
b(B,R
recovery,G
recovery),
Wherein, the G in the above-listed formula
Recovery, R
RecoveryWith B
Recovery: be respectively that this has revised green filtering parameter, this has revised red filtering parameter and this has revised the blue color filtered parameter; Filter
g, Filter
rWith Filter
b: be respectively green channel filter, red channel filter and blue channel filter.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009101336165A CN101854468B (en) | 2009-04-02 | 2009-04-02 | Repairing method for removing image noise |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2009101336165A CN101854468B (en) | 2009-04-02 | 2009-04-02 | Repairing method for removing image noise |
Publications (2)
Publication Number | Publication Date |
---|---|
CN101854468A CN101854468A (en) | 2010-10-06 |
CN101854468B true CN101854468B (en) | 2011-10-19 |
Family
ID=42805713
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2009101336165A Expired - Fee Related CN101854468B (en) | 2009-04-02 | 2009-04-02 | Repairing method for removing image noise |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101854468B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI528815B (en) * | 2013-02-04 | 2016-04-01 | 鈺立微電子股份有限公司 | Image processing method capable of reducing image noise |
CN107730454B (en) * | 2016-08-10 | 2021-02-05 | 苏州中科天启遥感科技有限公司 | Linear-tracking-based method for removing horizontal stripe noise in linear array image |
CN109510920B (en) * | 2017-09-14 | 2021-03-23 | 奇景光电股份有限公司 | Pixel interpolation circuit and pixel interpolation method |
CN112561776B (en) * | 2019-09-25 | 2023-08-22 | 杭州海康威视数字技术股份有限公司 | image processing method |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1457609A (en) * | 2001-02-06 | 2003-11-19 | 皇家菲利浦电子有限公司 | Contour filter formimage sensor output signals |
CN1666228A (en) * | 2002-07-04 | 2005-09-07 | 皇家飞利浦电子股份有限公司 | Method and apparatus for signal processing, computer program product, computing system and camera |
JP2006135564A (en) * | 2004-11-05 | 2006-05-25 | Casio Comput Co Ltd | Device and method for pixel interpolation |
CN1933550A (en) * | 2005-05-26 | 2007-03-21 | 智多微电子(上海)有限公司 | Bell field signal image processing method and module |
-
2009
- 2009-04-02 CN CN2009101336165A patent/CN101854468B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1457609A (en) * | 2001-02-06 | 2003-11-19 | 皇家菲利浦电子有限公司 | Contour filter formimage sensor output signals |
CN1666228A (en) * | 2002-07-04 | 2005-09-07 | 皇家飞利浦电子股份有限公司 | Method and apparatus for signal processing, computer program product, computing system and camera |
JP2006135564A (en) * | 2004-11-05 | 2006-05-25 | Casio Comput Co Ltd | Device and method for pixel interpolation |
CN1933550A (en) * | 2005-05-26 | 2007-03-21 | 智多微电子(上海)有限公司 | Bell field signal image processing method and module |
Also Published As
Publication number | Publication date |
---|---|
CN101854468A (en) | 2010-10-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102663719B (en) | Bayer-pattern CFA image demosaicking method based on non-local mean | |
JP5672776B2 (en) | Image processing apparatus, image processing method, and program | |
US8310567B2 (en) | Methods of modeling an integrated noise in an image sensor and methods of reducing noise using the same | |
US8605164B2 (en) | Image processing apparatus, control method therefor, and storage medium | |
US20030095717A1 (en) | Noise cleaning and interpolating sparsely populated color digital image using a variable noise cleaning kernel | |
CN105303530B (en) | Textile image moire fringes removing method based on low-rank sparse matrix decomposition | |
US7995868B2 (en) | Image processing method and image processing device | |
CN101854468B (en) | Repairing method for removing image noise | |
EP2557798A2 (en) | Apparatus and method for removing noise generated from image sensor | |
CN100568913C (en) | Edge crispening color interpolation method based on gradient | |
US20130051664A1 (en) | Image processing device | |
CN103907134B (en) | Image processing apparatus, image processing method and camera head | |
WO2013022113A1 (en) | Image processing apparatus and control method therefor | |
CN103237158B (en) | Image processing method, image acquisition system and image processor | |
JP2011171885A (en) | Image processing apparatus and image processing method | |
CN107248147B (en) | Method for eliminating dead pixel in RAW-RGB image | |
CN105260991A (en) | Adaptive hybrid noise reduction method for CMOS image sensor noise reduction processing | |
CN114902276A (en) | Image processor | |
US8218867B2 (en) | Compensation method for removing image noise | |
CN101489142B (en) | Colored interpolation method for digital image | |
CN105049820B (en) | IMAGE PROCESSING APPARATUS, IMAGING APPARATUS, and IMAGE PROCESSING METHOD | |
CN109788261A (en) | Color displacement bearing calibration and device | |
CN113840124A (en) | Image processing method and system | |
US9451222B2 (en) | Color processing of digital images | |
KR101994897B1 (en) | Demosaicing system for color imgae and demosaicing method therefor |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20111019 Termination date: 20150402 |
|
EXPY | Termination of patent right or utility model |