CN104125420A - Fixed-pattern noise removing method - Google Patents
Fixed-pattern noise removing method Download PDFInfo
- Publication number
- CN104125420A CN104125420A CN201310150703.8A CN201310150703A CN104125420A CN 104125420 A CN104125420 A CN 104125420A CN 201310150703 A CN201310150703 A CN 201310150703A CN 104125420 A CN104125420 A CN 104125420A
- Authority
- CN
- China
- Prior art keywords
- pixel
- hurdle
- photosensitive area
- offset
- pattern noise
- 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.)
- Granted
Links
Landscapes
- Facsimile Image Signal Circuits (AREA)
- Transforming Light Signals Into Electric Signals (AREA)
- Color Television Image Signal Generators (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a fixed-pattern noise removing method, which is used in an image sensor. The fixed-pattern noise removing method comprises steps of calculating each compensation value corresponding to each pixel column according to a plurality of compensation pixel values in each pixel column, and carrying out compensation when a plurality of photosensitive area pixel values in a photosensitive area are sampled according to each compensation value of each pixel column so as to generate a plurality of compensation photosensitive area pixel values, wherein the plurality of photosensitive area pixel values are photosensitively generated by the plurality of photosensitive area pixels in the photosensitive area.
Description
Technical field
The present invention relates to a kind of fixed pattern noise removal method, espespecially a kind of for image sensor, only need each offset that calculates corresponding each hurdle pixel to compensate, and can save storage space and reduce the fixed pattern noise removal method of computational complexity.
Background technology
In general, in image sensor (image sensor), because the pixel for sensing image is in size, spacing, efficiency inconsistent, and with a row sampler, as associated two sampler (correlation double sampling, CDS), sample by column photosensitive pixel value and have error; That is to say, one specific assignment sampling device is corresponding to particular bar pixel, when therefore in a row sampler, sampler is inhomogeneous in processing procedure and design, the pixel value of institute's acquisition hurdle pixel can cause error to have noise in image vertical mode because sampler is inhomogeneous, cause the image that image sensor captures can there is because of the existing disappearance of hardware structure nonrandom fixed pattern noise (fixed pattern noise, FPN).
For instance, please refer to Figure 1A and Figure 1B, Figure 1A and Figure 1B are respectively a real image IMG and have the schematic diagram of a noise image NIMG of hurdle fixed pattern noise.As shown in Figure 1A and Figure 1B, because specific assignment sampling device is inhomogeneous in processing procedure and design corresponding to sampler in particular bar pixel and a row sampler, so the pixel value of institute's acquisition hurdle pixel can have because sampler is inhomogeneous steady noise (vertical bar line as shown in Figure 1B) in image vertical mode.
In the case, known technology shines (black picture) by taking a scene with the shading of a same condition, then to subtracting each other, can eliminate fixed pattern noise by these two.Yet this mode need be noted down and store whole shading and control (temperature, yield value, time for exposure) according to doing different parameters, needs very large memory.In view of this, known technology has improved necessity in fact.
Summary of the invention
Therefore, it is a kind of for image sensor that main purpose of the present invention is to provide, and only needs each offset that calculates corresponding each hurdle pixel to compensate, and can save storage space and reduce the fixed pattern noise removal method of computational complexity.
The present invention discloses a kind of fixed pattern noise removal method, and for an image sensor, this fixed pattern noise removal method includes according to a plurality of compensation pixel values in each hurdle pixel, calculates each offset of corresponding this each hurdle pixel; And according to this each offset of this each hurdle pixel, when the pixel value of a plurality of photosensitive areas of sampling one photosensitive area, compensate, to produce a plurality of compensation photosensitive area pixel value; Wherein, in this photosensitive area, the pixel sensitization of a plurality of photosensitive areas produces the plurality of photosensitive area pixel value.
At this, coordinate detailed description and claims of following diagram, embodiment, by address other object of the present invention and advantage and be specified in after.
Accompanying drawing explanation
Figure 1A and Figure 1B are respectively a real image and have the schematic diagram of a noise image of hurdle fixed pattern noise.
Fig. 2 is the schematic diagram of the embodiment of the present invention one fixed pattern noise removal flow process.
Fig. 3 is the schematic diagram of the embodiment of the present invention one image sensor.
Fig. 4 is the schematic diagram of another image sensor of the embodiment of the present invention.
Fig. 5 is the schematic diagram of another image sensor of the embodiment of the present invention.
Fig. 6 to Fig. 8 is the schematic diagram of other three image sensors of the embodiment of the present invention.
Wherein, description of reference numerals is as follows:
20 flow processs
200,202,204,206 steps
30,40,50,60,70,80 image sensors
32 image processors
300,500 picture element matrixs
302 samplers
304,504 Memory Controllers
306,308,506,508 linear memories
310,312 multipliers
314 adders
316,402,316 ＇, 402 ＇ subtracters
318,318 ＇ analog-digital converters
320,520 photosensitive areas
322,522 shading regions
400 computing units
600,700,800 digital analog converters
IMG real image
NIMG noise image
CAPV
i, CAPV
i＇, CAPV
i＂ compensation photosensitive area pixel value
ID image data
R red pixel
Gr, Gb green pixel
B blue pixel
APV
iphotosensitive area pixel value
AG analog gain
OBA
ishading region certain primary color pixel average
OBCA
ihurdle, shading region pixel average
Embodiment
Please refer to Fig. 2, Fig. 2 is the schematic diagram of the embodiment of the present invention one fixed pattern noise removal (fixed pattern noise, FPN) flow process 20.As shown in Figure 2, fixed pattern noise removal flow process 20 is for an image sensor, and it comprises the following steps:
Step 200: start.
Step 202: according to a plurality of compensation pixel values in each hurdle pixel, calculate each offset of corresponding this each hurdle pixel.
Step 204: according to this each offset of this each hurdle pixel, compensate when the pixel value of a plurality of photosensitive areas of sampling one photosensitive area, to produce a plurality of compensation photosensitive area pixel value; Wherein, in this photosensitive area, the pixel sensitization of a plurality of photosensitive areas produces the plurality of photosensitive area pixel value.
Step 206: finish.
According to fixed pattern noise removal flow process 20, the present invention is first according to a plurality of compensation pixel values in each hurdle pixel, calculate each offset of corresponding this each hurdle pixel, again according to this each offset of this each hurdle pixel, when a plurality of photosensitive areas pixel sensitization in sampling one photosensitive area produces a plurality of photosensitive areas pixel value, the plurality of photosensitive area pixel value is compensated, to produce a plurality of compensation photosensitive area pixel value.In this case, because the present invention only need store each offset of corresponding this each hurdle pixel or even need not store in memory, therefore compared to known technology, need in memory, store whole shading according to can significantly saving storage space in memory.Thus, the present invention only needs each offset that calculates corresponding each hurdle pixel to compensate, and can save storage space and reduce computational complexity.
For instance, please refer to Fig. 3, Fig. 3 is the schematic diagram of the embodiment of the present invention one image sensor 30.As shown in Figure 3, image sensor 30 can produce compensation photosensitive area pixel value CAPV
igiving an image processor 32 makes it produce an image data ID, image sensor 30 includes a picture element matrix 300, one sampler 302, one Memory Controller 304, linear memory 306, 308, multiplier 310, 312, one adder 314, one subtracter 316 and an analog-digital converter (analog to digital converter, ADC) 318, wherein, picture element matrix 300 includes 320 and one shading region, a photosensitive area (Active Pixel Area) (Optical Black Area) 322, and picture element matrix 300 is Bel figure (Bayer Pattern) structure, therefore its each pixel is only a red pixel R, green pixel Gr, in the middle of Gb and a blue pixel B three primary colors pixel, one is with sensing particular color.
In simple terms, under an initial condition, it is a specific operation gain that image sensor 30 can first be set photosensitivity (ISO), make sampler 302 under this specific operation gain, sample and calculate 322Zhong Ge hurdle, a shading region pixel in each hurdle primary colors pixel average sum (CP of each former color pixel
(c) j)/CPN
(c) j, sum (CP
(c) k)/CPN
(c) keach primary colors pixel average AVE with this each former color pixel in shading region 322
j, AVE
kdifference LM
306 (c) j, LM
308 (c) k(the certain primary color pixel difference average and that whole certain primary color pixels are average of particular bar is the fixed pattern noise of the certain primary color pixel of particular bar), by Memory Controller 304, be stored in linear memory 306,308 again, (all not sensitization of pixel in shading region 322 as follows, can have in this embodiment 16 row in order to calculate average, and in 322Zhong Ge hurdle, shading region pixel, each primary colors pixel value is a plurality of compensation pixel values in each hurdle pixel described in step 202):
LM
306(c)j=AVE
j–sum(CP
(c)j)/CPN
(c)j,AVE
j=sum(OBP
j)/OBPN
j,j=R,Gr
LM
308(c)k=AVE
k–sum(CP
(c)k)/CPN
(c)k,AVE
k=sum(OBP
k)/OBPN
k,k=Gb,B
Wherein, sum (CP
(c) j)/CPN
(c) j, sum (CP
(c) k)/CPN
(c) krepresent that respectively the pixel value summation of a field c Central Plains color pixel j, k is divided by the quantity of former color pixel j, k, sum (OBP
j)/OBPN
j, sum (OBP
k)/OBPN
krepresent that respectively the pixel value summation of shading region 322 Central Plains color pixel j, k is divided by the quantity of former color pixel j, k.
For instance, left side the 1st hurdle Yin Beier graph structure is only had red pixel R and a green pixel Gb, when wish is calculated the hurdle red pixel mean value sum (CP on the 1st hurdle
(1) R)/CPN
(1) Rtime, the pixel value of all red pixel R in the 1st hurdle in shading region 322 can be added then divided by the quantity of the 1st hurdle red pixel R, then calculate again the red pixel mean value AVE of red pixel R in shading region 322
r, the pixel value that is about to all red pixel R in shading region 322 is added then divided by the quantity of all red pixel R, then subtracts each other and gets difference LM
306 (1) Rbe stored in the position of corresponding the 1st hurdle red pixel R.The rest may be inferred, can calculate the difference LM of corresponding red pixel R, green pixel Gr, Gb and blue pixel B in other hurdle
306 (c) j, LM
308 (c) kbe stored in respectively again in linear memory 306,308, wherein, owing to being listed as separately in Bel's graph structure, there is same pixel arrangement, therefore only need two alignment memories 306,308 can store the difference of the mean value of the former color pixel in all hurdles, and because fixed pattern noise is steady noise, therefore can set specific operation gain is that a maximum actual gain is clearly to note down.
Then, at initial condition image sensor 30, setting photosensitivity is specific operation gain, and then Memory Controller 304 is by difference LM
306 (c) j, LM
308 (c) kbe stored in after linear memory 306,308, under a sensitization state, 320 samplings of 302 pairs of photosensitive areas of sampler produce photosensitive area pixel value APV
i, and image sensor 30 set photosensitivity be an analog gain AG by multiplier 310 according to this to photosensitive area pixel value APV
igain, then photosensitive area pixel value APV after being gained by analog-digital converter 318
itransfer a digital form to.At the same time, Memory Controller 304 can read corresponding difference LM in linear memory 306,308
306 (c) j, LM
308 (c) ka ratio that is multiplied by by multiplier 312 the specific operation gain that analog gain AG and initial condition set again obtains fixed pattern noise under analog gain AG (because fixed pattern noise is steady noise, therefore hypothesis can be amplified and dwindle with gain equal proportion), then add up under analog gain AG in shading region 322 all shading region certain primary color pixel average OBA of certain primary color pixels by adder 314
i(shading region 322 is irradiation not, so shading region certain primary color pixel average OBA
irepresent the not size of existing dark current under irradiation of certain primary color pixel) obtain this each offset of each hurdle pixel, then subtracter 316 is again by photosensitive area pixel value APV
ithis each offset that deducts corresponding each hurdle pixel is compensated photosensitive area pixel value CAPV
i, as follows:
OBA
i=sum(AGOBP
i)/AGOBPN
i
CAPV
i=APV
i-OBA
i+LM
(c)i*AG
i=R,Gr,R,Gr；LM
(c)i=LM
306(c)j、LM
308(c)k
Wherein, sum (AGOBP
j)/AGOBPN
jrepresent that the pixel value summation of shading region 322 Central Plains color pixel i under analog gain AG is divided by the quantity of former color pixel i.
Thus, image sensor 30 only need utilize two alignment memories 306,308 to store difference LM in initial condition
306 (c) j, LM
308 (c) k, then at sensitization state, the simple combination by multiplier 312, adder 314 and subtracter 316 can compensate and eliminates fixed pattern noise, therefore can save storage space and reduce computational complexity.
Each offset that can calculate corresponding each hurdle pixel that is mainly that it should be noted that above-described embodiment compensates, to save storage space and to reduce computational complexity.Those of ordinary skills work as and can modify according to this or change, and are not limited to this.For instance, please refer to Fig. 4, Fig. 4 is the schematic diagram of another image sensor 40 of the embodiment of the present invention.Image sensor 40 is similar to image sensor 30 parts, so identical assembly and the signal of function represents with same-sign.Image sensor 40 is with the main difference of image sensor 30, image sensor 40 under initial condition by memory storage difference, but 320 samplings of 302 pairs of photosensitive areas of sampler produce photosensitive area pixel value APV under sensitization state
iand when multiplier 310 carries out subsequent operation with analog-digital converter 318, by a computing unit 400, sample and calculate simultaneously each hurdle, shading region pixel average OBCA of 322Zhong Ge hurdle, shading region pixel
ias each offset of each hurdle pixel, (this hurdle pixel average comprises dark current composition and fixed pattern noise component simultaneously, and in 322Zhong Ge hurdle, shading region pixel, each hurdle pixel value is a plurality of compensation pixel values in each hurdle pixel described in step 202), then by subtracter 402 by photosensitive area pixel value APV
ithis each offset that deducts corresponding each hurdle pixel (is each hurdle, shading region pixel average OBCA
i) be compensated photosensitive area pixel value CAPV
i＇, as follows:
OBCA
i=sum(OBCP
i)/OBCPN
i
CAPV
i＇=APV
i-OBCA
i
i=R,Gr,R,Gr
Wherein, sum (OBCP
j)/OBCPN
jrepresent under analog gain AG in shading region 322 that the pixel value summation of specific fields Central Plains color pixel i is divided by the quantity of former color pixel i.
Thus, due to image sensor 40 direct sample calculate each hurdle, shading region pixel average OBCA of 322Zhong Ge hurdle, shading region pixel when 320 sensitization of photosensitive area
ias each offset of each hurdle pixel, then eliminate fixed pattern noise by can compensating of subtracter 402, therefore do not need memory and can reduce computational complexity.
Moreover, please refer to Fig. 5, Fig. 5 is the schematic diagram of another image sensor 50 of the embodiment of the present invention.Image sensor 50 is similar to image sensor 30 parts, so identical assembly and the signal of function represents with same-sign.Image sensor 50 is with the main difference of image sensor 30, it is after a specific operation gains that image sensor 50 is set photosensitivity under an initial condition, the link that can cut off 520Zhong photosensitive area, a photosensitive area pixel of a sampler 502 and a picture element matrix 500 (links owing to disconnecting between sampler 502 and picture element matrix 500, therefore 502 read values of sampler are the circuit values of corresponding picture element matrix 500 parts), then calculate sampler 502 corresponding to each hurdle primary colors circuit mean value sum (CC of each former color pixel in 520Zhong Ge hurdle, photosensitive area pixel
(c) j)/CCN
(c) j, sum (CC
(c) k)/CCN
(c) kwith each the primary colors circuit mean value CAVE of sampler 502 corresponding to this each former color pixel in photosensitive area 520
j, CAVE
kdifference LM
506 (c) j, LM
508 (c) k(the average difference of circuit values average corresponding to the circuit values of the certain primary color pixel of particular bar and corresponding to whole certain primary color pixels is the fixed pattern noise that sampler 502 causes corresponding to the certain primary color pixel of particular bar), by a Memory Controller 504, be stored in linear memory 506,508 following (in 520Zhong Ge hurdle, photosensitive area pixel, each primary colors circuit mean value is a plurality of compensation pixel values in each hurdle pixel described in step 202):
LM
506(c)j=CAVE
j–sum(CC
(c)j)/CCN
(c)j
CAVE
j=sum(APC
j)/APCN
j,j=R,Gr
LM
508(c)k=CAVE
k–sum(CC
(c)k)/CCN
(c)k
CAVE
k=sum(APC
k)/APCN
k,k=R,Gr
Wherein, sum (CC
(c) j)/CCN
(c) j, sum (CC
(c) k)/CCN
(c) krepresent respectively sampler 502 corresponding to the circuit values summation of photosensitive area 520 1 field c Central Plains color pixel j, k the quantity divided by former color pixel j, k, sum (APC
j)/APCN
j, sum (APC
k)/APCN
krepresent respectively sampler 502 corresponding to the circuit values summation of photosensitive area 520 Central Plains color pixel j, k the quantity divided by former color pixel j, k.
For instance, left side the 1st hurdle Yin Beier graph structure is only had red pixel R and a green pixel Gb, when wish is calculated the hurdle red circuit mean value sum (CC on the 1st hurdle
(1) R)/CCN
(1) Rtime, can first will between sampler 502 and picture element matrix 500, disconnect and linking, then sampler 502 simulation normal runnings link in the 1st hurdles, photosensitive area 520 each red pixel R and obtain corresponding circuit values and add up then divided by the quantity of the 1st hurdle red pixel R, then calculate the red circuit mean value CAVE corresponding to red pixel R in photosensitive area 520 again
r, i.e. sampler 502 simulation normal runnings link in photosensitive areas 520 all red pixel R and obtain corresponding circuit values and add up then divided by the quantity of all red pixel R, then subtract each other and get difference LM
506 (1) Rbe stored in the position of corresponding the 1st hurdle red pixel R.The rest may be inferred, can calculate the difference LM of corresponding red pixel R, green pixel Gr, Gb and blue pixel B in other hurdle
506 (c) j, LM
508 (c) kbe stored in respectively again in linear memory 506,508, wherein, owing to being listed as separately in Bel's graph structure, there is same pixel arrangement, therefore only need two alignment memories 506,508 can store the difference of the mean value of the former color pixel in all hurdles, and because fixed pattern noise is steady noise, therefore can set specific operation gain is that a maximum actual gain is clearly to note down.
Then, at initial condition image sensor 50, setting photosensitivity is that specific operation gains and disconnects link between sampler 502 and picture element matrix 500, and then Memory Controller 504 is by difference LM
506 (c) j, LM
508 (c) kbe stored in after linear memory 306,308, under a sensitization state, image sensor 50 links between conducting sampler 502 and picture element matrix 500 again, makes 520 samplings of 502 pairs of photosensitive areas of sampler produce photosensitive area pixel value APV
i, and image sensor 50 set photosensitivity be analog gain AG by multiplier 310 according to this to photosensitive area pixel value APV
igain, then photosensitive area pixel value APV after being gained by analog-digital converter 318
itransfer a digital form to.At the same time, Memory Controller 504 can read corresponding difference LM in linear memory 506,508
506 (c) j, LM
508 (c) ka ratio that is multiplied by by multiplier 312 the specific operation gain that analog gain AG and initial condition set again obtains fixed pattern noise under analog gain AG (because fixed pattern noise is steady noise, therefore hypothesis can be amplified and dwindle with gain equal proportion), then add up under analog gain AG in shading region 322 all shading region certain primary color pixel average OBA of certain primary color pixels by adder 314
i(shading region 322 is irradiation not, so shading region certain primary color pixel average OBA
irepresent the not size of existing dark current under irradiation of certain primary color pixel) obtain this each offset of each hurdle pixel, then subtracter 316 is again by photosensitive area pixel value APV
ithis each offset that deducts corresponding each hurdle pixel is compensated photosensitive area pixel value CAPV
i＂, as follows:
OBA
i=sum(AGOBP
i)/AGOBPN
i
CAPV
i＂=APV
i-OBA
i+LM
(c)i*AG
i=R,Gr,R,Gr；LM
(c)i=LM
506(c)j、LM
508(c)k
Wherein, sum (AGOBP
j)/AGOBPN
jrepresent that the pixel value summation of shading region 322 Central Plains color pixel i under analog gain AG is divided by the quantity of former color pixel i.
Thus, image sensor 50 also only need utilize two alignment memories 506,508 to store difference LM in initial condition
506 (c) j, LM
508 (c) k, under sensitization state, pass through again multiplier 312, the simple combination of adder 314 and subtracter 316 can compensate eliminates fixed pattern noise, therefore can save storage space and reduce computational complexity, and because disconnecting between sampler 502 and picture element matrix 500, image sensor 50 links, each offset of each each hurdle pixel of primary colors circuit mean value calculation in recycling 520Zhong Ge hurdle, photosensitive area pixel, therefore compared to image sensor 30, can have less shading region 522(image sensor 50 does not need to utilize shading region 522 to average with offset value calculation, therefore can only there are 4 row).
In addition, in the above-described embodiments, image sensor 30,40,50 all under digital form to each offset and photosensitive area pixel value APV
icompensate (is that each offset is digital form, at photosensitive area pixel value APV
iafter being treated to digital form, analog-digital converter 318 just compensates).Yet, in other embodiments, also can be to each offset and photosensitive area pixel value APV under analog form
icompensate.Please refer to Fig. 6 to Fig. 8, Fig. 6 to Fig. 8 is the schematic diagram of other three image sensors 60,70,80 of the embodiment of the present invention.Image sensor 60,70,80 is similar to image sensor 30,40,50 parts, so identical assembly and the signal of function represents with same-sign.The main difference of image sensor 60,70,80 and image sensor 30,40,50 be image sensor 60,70,80 first respectively by each offset through digital analog converter (digital to analog converter, DAC) 600,700,800 by digital form, be converted to after analog form, then by subtracter 316 ＇, 402 ＇, after gaining, be still the photosensitive area pixel value APV of analog form respectively
iunder analog form, compensate, then be converted to digital form by analog-digital converter 318 ＇.Thus, image sensor 60,70,80 can be to each offset and photosensitive area pixel value APV under analog form
icompensate.
In known technology, need to shine (black picture) by taking a scene with the shading of a same condition, by these two, to subtracting each other, can eliminate fixed pattern noise again, but this mode need be noted down and store whole shading and control (temperature according to doing different parameters, yield value, time for exposure), need very large memory.In comparison, each offset that the present invention can calculate corresponding each hurdle pixel compensates, therefore only need in memory, store each offset of corresponding this each hurdle pixel or even need in memory, not store, and can save storage space and reduce computational complexity.
The foregoing is only the preferred embodiments of the present invention, be not limited to the present invention, for a person skilled in the art, the present invention can have various modifications and variations.Within the spirit and principles in the present invention all, any modification of doing, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.
Claims (10)
1. a fixed pattern noise removal method, for an image sensor, is characterized in that, includes:
According to a plurality of compensation pixel values in each hurdle pixel, calculate each offset of corresponding this each hurdle pixel;
And
According to this each offset of this each hurdle pixel, when the pixel value of a plurality of photosensitive areas of sampling one photosensitive area, compensate, to produce a plurality of compensation photosensitive area pixel value;
Wherein, in this photosensitive area, the pixel sensitization of a plurality of photosensitive areas produces the plurality of photosensitive area pixel value.
2. fixed pattern noise removal method as claimed in claim 1, is characterized in that, according to the plurality of compensation pixel value in this each hurdle pixel, the step of calculating this each offset of corresponding this each hurdle pixel includes:
Set a specific operation gain; And
Under this specific operation gain and an initial condition, sample and calculate in each hurdle primary colors pixel average of each former color pixel in Zhong Gaige hurdle, a shading region pixel and shading region the difference of each primary colors pixel average of this each former color pixel, and be stored in a memory;
Wherein, all not sensitization of pixel in this shading region.
3. fixed pattern noise removal method as claimed in claim 2, is characterized in that, according to the plurality of compensation pixel value in this each hurdle pixel, the step of calculating this each offset of corresponding this each hurdle pixel includes:
Calculate under an analog gain in this shading region all shading region certain primary color pixel average of certain primary color pixels; And
By in this each primary colors pixel average of this each former color pixel in this each hurdle pixel and this each hurdle pixel all the difference of this each pixel average of pixels be multiplied by the ratio that this analog gain and this specific operation gain, then add up this shading region pixel average as this each offset of this each hurdle pixel.
4. fixed pattern noise removal method as claimed in claim 2, is characterized in that, this specific operation gain is a maximum actual gain.
5. fixed pattern noise removal method as claimed in claim 1, is characterized in that, according to the plurality of compensation pixel value in this each hurdle pixel, the step of calculating this each offset of corresponding this each hurdle pixel includes:
In this photosensitive area, during the pixel sensitization of the plurality of photosensitive area, sample and calculate each hurdle, shading region pixel average of Zhong Gaige hurdle, a shading region pixel as each offset of this each hurdle pixel.
6. fixed pattern noise removal method as claimed in claim 1, is characterized in that, according to the plurality of compensation pixel value in this each hurdle pixel, the step of calculating this each offset of corresponding this each hurdle pixel includes:
Set a specific operation gain;
Cut off the link of the plurality of photosensitive area pixel in a sampler and this photosensitive area; And
Under this specific operation gain and an initial condition, calculate this sampler corresponding to each hurdle primary colors circuit mean value of each former color pixel in Zhong Gaige hurdle, this photosensitive area pixel and this sampler the difference corresponding to each primary colors circuit mean value of this each former color pixel in this photosensitive area, and be stored in a memory.
7. fixed pattern noise removal method as claimed in claim 6, is characterized in that, according to the plurality of compensation pixel value in this each hurdle pixel, the step of calculating this each offset of corresponding this each hurdle pixel includes:
Calculate a shading region pixel average of whole pixels in a shading region; And
This sampler is multiplied by a ratio of an analog gain and the gain of this specific operation corresponding to each primary colors circuit mean value and this sampler of each former color pixel in Zhong Gaige hurdle, this photosensitive area pixel corresponding to the difference of each hurdle circuit mean value of whole pixels in this each hurdle pixel, then adds up this shading region pixel average as this each offset of this each hurdle pixel;
Wherein, all not sensitization of pixel in this shading region.
8. fixed pattern noise removal method as claimed in claim 6, is characterized in that, this specific operation gain is a maximum actual gain.
9. fixed pattern noise removal method as claimed in claim 1, is characterized in that, this each offset and the plurality of photosensitive area pixel value compensate under a digital form.
10. fixed pattern noise removal method as claimed in claim 1, is characterized in that, this each offset and the plurality of photosensitive area pixel value compensate under an analog form.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310150703.8A CN104125420B (en) | 2013-04-26 | 2013-04-26 | fixed pattern noise removal method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310150703.8A CN104125420B (en) | 2013-04-26 | 2013-04-26 | fixed pattern noise removal method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104125420A true CN104125420A (en) | 2014-10-29 |
| CN104125420B CN104125420B (en) | 2019-01-04 |
Family
ID=51770670
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310150703.8A Active CN104125420B (en) | 2013-04-26 | 2013-04-26 | fixed pattern noise removal method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104125420B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107820618A (en) * | 2017-09-30 | 2018-03-20 | 深圳市汇顶科技股份有限公司 | Sensing pixels unit and optical fingerprint sensor |
| CN110365925A (en) * | 2019-07-11 | 2019-10-22 | 北京安酷智芯科技有限公司 | A kind of reading circuit, biasing circuit noise determining method and device |
| CN113365002A (en) * | 2021-06-04 | 2021-09-07 | 深圳市汇春科技股份有限公司 | Fixed mode noise dynamic compensation device and method and terminal equipment |
| CN114079735A (en) * | 2020-08-19 | 2022-02-22 | 瑞昱半导体股份有限公司 | Image compensation system for fixed image noise |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1770829A (en) * | 2004-11-02 | 2006-05-10 | 索尼株式会社 | Signal processing apparatus and signal processing method for solid-state image pickup element and image pickup apparatus |
| CN1783957A (en) * | 2004-11-26 | 2006-06-07 | 株式会社东芝 | Solid-state imaging apparatus |
| EP1993277A2 (en) * | 2007-05-15 | 2008-11-19 | Sony Corporation | Image pickup apparatus and method of correcting captured image data |
| CN101909144A (en) * | 2009-06-08 | 2010-12-08 | 华晶科技股份有限公司 | Image data correction method |
| US20110128420A1 (en) * | 2009-11-30 | 2011-06-02 | Stmicroelectronics (Research & Development) Limited | Method and system for repeated fixed pattern noise correction |
-
2013
- 2013-04-26 CN CN201310150703.8A patent/CN104125420B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1770829A (en) * | 2004-11-02 | 2006-05-10 | 索尼株式会社 | Signal processing apparatus and signal processing method for solid-state image pickup element and image pickup apparatus |
| CN1783957A (en) * | 2004-11-26 | 2006-06-07 | 株式会社东芝 | Solid-state imaging apparatus |
| EP1993277A2 (en) * | 2007-05-15 | 2008-11-19 | Sony Corporation | Image pickup apparatus and method of correcting captured image data |
| CN101909144A (en) * | 2009-06-08 | 2010-12-08 | 华晶科技股份有限公司 | Image data correction method |
| US20110128420A1 (en) * | 2009-11-30 | 2011-06-02 | Stmicroelectronics (Research & Development) Limited | Method and system for repeated fixed pattern noise correction |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107820618A (en) * | 2017-09-30 | 2018-03-20 | 深圳市汇顶科技股份有限公司 | Sensing pixels unit and optical fingerprint sensor |
| CN110365925A (en) * | 2019-07-11 | 2019-10-22 | 北京安酷智芯科技有限公司 | A kind of reading circuit, biasing circuit noise determining method and device |
| CN110365925B (en) * | 2019-07-11 | 2021-08-31 | 北京安酷智芯科技有限公司 | Reading circuit, and method and device for determining bias circuit noise |
| CN114079735A (en) * | 2020-08-19 | 2022-02-22 | 瑞昱半导体股份有限公司 | Image compensation system for fixed image noise |
| CN114079735B (en) * | 2020-08-19 | 2024-02-23 | 瑞昱半导体股份有限公司 | Image compensation system for fixed image noise |
| CN113365002A (en) * | 2021-06-04 | 2021-09-07 | 深圳市汇春科技股份有限公司 | Fixed mode noise dynamic compensation device and method and terminal equipment |
| CN113365002B (en) * | 2021-06-04 | 2022-09-16 | 深圳市汇春科技股份有限公司 | Fixed mode noise dynamic compensation device and method and terminal equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104125420B (en) | 2019-01-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10136107B2 (en) | Imaging systems with visible light sensitive pixels and infrared light sensitive pixels | |
| US9148594B2 (en) | Fixed pattern noise removal method | |
| JP6096243B2 (en) | Image data processing method and system | |
| US8160381B2 (en) | Method and apparatus for image noise reduction using noise models | |
| JP4960309B2 (en) | Solid-state imaging device | |
| US11350063B2 (en) | Circuit for correcting lateral chromatic abberation | |
| CN104125420A (en) | Fixed-pattern noise removing method | |
| US20140375861A1 (en) | Image generating apparatus and method | |
| EP3476117A1 (en) | Fixed pattern noise mitigation for a thermal imaging system | |
| KR102406215B1 (en) | Methods of correcting color fringe and processing image data using the same | |
| WO2007085004A2 (en) | Hand jitter reduction for compensating for linear displacement | |
| US9628730B2 (en) | Dark current gradient estimation using optically black pixels | |
| US11763421B2 (en) | Circuit for combined down sampling and correction of image data | |
| KR20170095572A (en) | Image sensor and method for generating restoration image | |
| TW200806048A (en) | Method and apparatus for gray value identification for white balance | |
| CN102811320A (en) | Black order compensating circuit, image sensor and relevant method | |
| TWI422224B (en) | Black level compensation circuit, image sensor and associated method | |
| US9066056B2 (en) | Systems for constant hue and adaptive color correction image processing | |
| CN105357452B (en) | The method of the adaptable reduction power consumption of image sensor | |
| KR100801057B1 (en) | CMOS image sensor having color correction block and method there-of | |
| US11037272B2 (en) | Reduction of line banding image artifacts | |
| US20240029308A1 (en) | Apparatus and method for performing color transformation on raw sensor images | |
| CN117082324A (en) | Photosensitive chip, image processing method and electronic equipment | |
| TWI601427B (en) | Imaging systems with clear filter pixels | |
| TW201436579A (en) | Imaging systems with clear filter pixels |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |