CN1992792A - Dynamic defect correction method and apparatus, and electronic equipment thereof - Google Patents
Dynamic defect correction method and apparatus, and electronic equipment thereof Download PDFInfo
- Publication number
- CN1992792A CN1992792A CNA2005100488656A CN200510048865A CN1992792A CN 1992792 A CN1992792 A CN 1992792A CN A2005100488656 A CNA2005100488656 A CN A2005100488656A CN 200510048865 A CN200510048865 A CN 200510048865A CN 1992792 A CN1992792 A CN 1992792A
- Authority
- CN
- China
- Prior art keywords
- value
- signal
- pixel
- signal strength
- pending
- 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
Images
Landscapes
- Transforming Light Signals Into Electric Signals (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a one-dimension defect correct method and relative device, wherein based on the one-dimension array of pixel, it judges the local maximum signal strength, minimum signal strength and the average value of nearby regions, judges if the pixel is defect one. When the pixel is needed to correct, find the method.
Description
Technical field
The invention relates to the bearing calibration and the device of picture element flaw in the correcting imaging device dynamically and the electronic installation that uses it, particularly, about using the method for correcting pixel defect and the device of one dimension defective deduction, and the electronic equipment with it.
Background technology
In recent years, electronic type photography or picture pick-up device are widely popular, and for example, these equipment comprise digital camera, digital code camera, have mobile phone of shoot function or the like.In these electronic type photographies or picture pick-up device, all has the electronic type imager usually.The signal that imager is obtained can be passed to image-processing circuit, device or chip, to carry out further image processing, and when for example sending the image processing chip to, need proofread and correct earlier the picture element flaw in the imager, this is the critical part of signal of video signal processing.
Generally speaking, imager comprises charge-coupled device (CCD) imager or complementary metal oxide semiconductors (CMOS) (CMOS) imager, can have some defect pixels in these imagers.The pattern of defective has a variety of, for example, be fixed in a bit, be fixed in zero point, sensitivity is high, sensitivity is extremely low, or the like.A kind of method that is called static defect correction is to set up the address table to write down these defect pixels and according to address it to be compensated.These address are to be recorded in the memory when the treatment for correcting of factory.
The defect pixel of imager can be more obvious because of the heat of device and problem such as aging, so, need to proofread and correct.One simple bearing calibration be apply median filter with when big the variation taken place with its filtering.But the critical value of median filter but is difficult to set.If the critical value height, then can't filtering most of defective, if critical value is low, then the high frequency of image partly will be judged by accident and median filter will be easy to generate false color.Though the method can extend to two-dimensional structure, erroneous judgement still can take place.This type of known technology is found in U.S. Patent number 6181830 and 6947083 and has disclosed the imager structure with static state and dynamic defect correction method.
Be published in " Fast Method for Noise Level Estimation and Denosing " literary composition of being shown by A.Bosco, A.Bruna, G.Messina, G.Spaminato etc. on the IEEE Transaction on ConsumerElectronics in January, 2005, disclose the noise that uses one of static mode method to comprise the dynamic defect pixel with reduction.Though the method can obtain the favorable noise filtering,, hardware cost and computation complexity are higher.
U.S. Patent number 6940549 discloses low-cost dynamic defect correction method.The method attempts pixel region is divided into high frequency image or low frequency image.Vacation is high frequency, then applies median filter.Though the method can be avoided some erroneous judgement of simple median filter,, still need critical value.In addition, can cause as the great bodily injury the high frequency image as known median filter, the method also can to high frequency partly original good image cause extensive damage.
Therefore, need a kind of dynamic method of correcting pixel defect or device,, can avoid again the high frequency image is caused great bodily injury so that can proofread and correct low frequency picture element flaw partly effectively.
Summary of the invention
Take into account the problems referred to above, the invention provides the dynamically method or the device of the defective of correcting image data.
The method according to this invention according to the one-dimensional array of pixel, is judged the mean value of local area maximum signal, local area minimum signal strength and adjacent domain, judges whether pixel is defect pixel.If pixel will be corrected, then judge how will be worth correction, but not as the known median filter, only average to proximity.
The method according to this invention does not need to increase line buffer or complex mathematical.In addition, the present invention can implement with software or hardware.
According to an aspect of the present invention, the image data defect correctin device is provided, comprise two low pass filters, a median filter, a maximum value detector, a minimum detector, gradually layer calculator, and minority glue logic mechanism.
Description of drawings
Fig. 1 is a flow chart, shows the embodiment according to dynamic pixel defect correction method of the present invention.
Fig. 2 is a calcspar, shows the embodiment according to dynamic pixel defect correctin device of the present invention.
Fig. 3 a and 3b are the performance measurement figure that shows respectively according to embodiments of the invention and known median filter.
Fig. 4 a and 4b are the another performance test figure that shows respectively according to embodiments of the invention and known median filter.
Embodiment
At first, under principle of the present invention is described.
According to the present invention, at first, judge that whether pixel signal intensities is greater than local area maximum signal (hereinafter to be referred as the local area maximum) or less than local area minimum signal strength (hereinafter to be referred as the local area minimum value).If, then this pixel is assumed to the visual defects pixel that is not to be selected for not.Other situation then is considered as this pixel visual defects pixel to be selected.Secondly, if pixel signal intensities, is then checked poor between the signal strength signal intensity of median filter output and the local area minimum value greater than the local area maximum; If pixel intensity, is then checked poor between median filter output and the local area maximum less than the local area minimum value.If difference is little, then this pixel is considered as defect pixel.Other situation is considered as pixel signal intensities the peak value of high frequency region.Produce one gradually layer parameter to represent whether this pixel is the confidence degree of defect pixel.When high confidence is spent, be output as median filter output; When low confidence was spent, output was local area maximum or the local area minimum value that keeps the original pixel intensity information of major part.According to the present invention, the advantage that has is for to carry out more defect correction at low frequency range, and carries out less modification at high frequency region.In addition, when high frequency region, the erroneous judgement probability is low, that is, can reserved high-frequency information, and naked eyes almost can't be differentiated defective.At low frequency partly, the efficient of defect correction is the same with median filter or better.
Then, will be with reference to the accompanying drawings, to illustrate according to embodiments of the invention.
(dynamic pixel defect correction method)
Fig. 1 is a flow chart, and the embodiment according to dynamic pixel defect correction method of the present invention is described.At first, with the signal strength signal intensity of present pixel to be processed (below, be called for short current pixel) represent with P (i), the signal strength signal intensity of two pixels before the current pixel in the same channel is represented with P (i-4) and P (i-2) respectively, and, the signal strength signal intensity of two pixels after current pixel that continue in co-channel is represented with P (i+2) and P (i+4) respectively.
As shown in Figure 1, signal strength signal intensity P (i) at step 102 input current pixel, then, find out maximum MaxProfile (i) and minimum M inProfile (i) among current pixel intensity P (i) and neighborhood pixels intensity P (i-4), P (i-2), P (i+2), the P (i+4), make these maximums MaxProfile and minimum M inProfile pass through low-pass filtering treatment, then, obtain two curves, one of them represents local area maximum MaxProfile, another curve is represented the local area minimum value, is called MinProfile.Share if consider hardware, then can calculate median filter output MedOutput simultaneously, account form is median filter output MedOutput=(P (i-2)+P (i+2))/2.
Then, judge that whether P (i) is greater than MaxProfile or less than MinProfile.If all do not meet this two condition, that is, P (i) promptly not greater than MaxProfile also not less than MinProfile, then need not proofread and correct, output P (i).This judgment processing at first judges that in step 104 whether P (i) is less than MinProfile (i) as shown in Figure 1.If for not, then proceed to step 108, judge that whether P (i) is greater than MaxProfile (i).If for not, then proceed to step 114, P (i) be used as output valve output.
If the judged result of step 104 is for being, that is P (i) then proceeds to step 106 less than MinProfile, check the difference between MaxProfile and the MedOutput, this difference is normalized to 0 to 1 scope, and is referred to as k.Then, proceed to step 110, obtain output valve Output=(1-k) * MedOutput (i)+k*MinProfile (i), and with its output.According to above-mentioned steps, know that as can be known parameter k can control the gradually layer of MedOutput and MinProfile.Gradually layer result is output.
If the judged result of step 108 be P (i) greater than MaxProfile, then processing can proceed to step 112, checks the difference between MinProfile and the MedOutput.If difference is little, representing local area is low frequency, and P (i) is a defect pixel.If difference is big, representing local area is high frequency, and P (i) may be for defect pixel or is not defect pixel.Herein, the judgment standard of difference size is that the ratio with the current pixel signal strength signal intensity is a fiducial value, if difference is greater than this fiducial value, then difference is big, if, be less than or equal to this fiducial value then difference be little.For example, this fiducial value can multiply by 1/5,1/10 or 1/20 for the current pixel signal strength signal intensity.Perhaps, if current pixel signal strength signal intensity during with the value representation between the 0-255, then this fiducial value is preferably in 4 to 18 scope, but not as limit.This difference is for judging the gradually layer parameter k of MedOutput and MaxProfile.Then, handle and proceed to step 116, obtain output valve Output=(1-k) * MedOutput (i)+k*MaxProfile (i), and with its output.According to above-mentioned steps, know that as can be known parameter k can control the gradually layer of MedOutput and MaxProfile.Gradually layer result is output.
According to embodiments of the invention, classification of defects is following three kinds of situations: situation 1, " zero defect "; Situation 2, " possible bright defective "; Situation 3, " possible dark defective ".In situation 1, output is identical with input.In situation 2, output is to export the peaked scope of local area at median filter.In situation 3, output is the scope that exports the local area minimum value at median filter to.The dynamic pixel defect correctin device
Then, will dynamic defect means for correcting 200 according to an embodiment of the invention be described with reference to figure 2.
As shown in Figure 2, by 4 delay elements 202,204,206,208, obtain 5 continuous pixel signal intensities data, P (i-4), P (i-2), P (i), P (i+2), P (i+4).The code name explanation of P (i-4), P (i-2), P (i), P (i+2), P (i+4) repeats no more as mentioned above herein.
In these pixel signal intensities, P (i-2) and P (i+2) can be fed to median filter 210, and P (i-4), P (i-2), P (i+2), P (i+4) can present simultaneously to maximum and find device 220, reach minimum value discovery device 230.And present signal strength values P (i) can be fed to the input B of comparator 240.
Median filter 210 can be average with P (i-2) and P (i+2), and (((P (i-2)+P (i+2))/2) also export mean value MedOutput to the input A of calculation element 270.The structure and the operation of calculation element 270 will be described in detail in the following.
Find can find out maximum, and it is fed to low pass filter 222 in the device 220 in maximum, then, output local area maximum MaxProfile.
Find can find out minimum value, and it is fed to low pass filter 232 in the device 230 in minimum value, then, output local area minimum M inProfile.
Then, comparator 250, local area maximum MaxProfile, the local area minimum M inProfile that can will input to it compare with P (i), and send comparative result to multiplexer 260,262,264 respectively, with according to comparative result, determine those data will be sent to parameter calculation apparatus 270 (in down, abbreviate Calc k as), and, determine which data will send gradually bed device 280 to, which data is for directly exporting multiplexer 264 to as final dateout output.
Shown in Fig. 2 below, the signal subtraction that parameter calculation apparatus 270 can be imported input A and B respectively, take absolute value again, this absolute difference is handled through the value of cutting, any absolute difference greater than a critical value (for example, value in 1/20 to 1/10 of aforesaid current pixel signal strength values or the arbitrary value among the 4-18) is considered as maximum without exception, at last, impose normalized again, then, export the result to gradually bed device 280 as the k value.
Then, the gradually operation of bed device 280 is described.As shown in Figure 2, gradually the bed device 280 signal MedOutput that can will input to input A multiply by (1-k), then, according to the output of comparator 240 being multiplied by k from the signal (MaxProfile or MinProfile) of B input input, then, with this two products addition, for example, (MedOutput* (1-k))+(MaxProfile*k), or (MedOutput* (1-k))+(MinProfile*k)).
Fig. 3 a shows effect according to an embodiment of the invention, and Fig. 3 b is the performance according to the median filter of known technology.In Fig. 3 a, dotted line L2 represents input signal, and solid line L1 is an output signal.The left portion of input signal is represented high-frequency information, and some defective in the low frequency information is partly represented on the right.Whether median filter needs a critical value to act on the decision correction.But in fact, known median filter is when making a lot of erroneous decisions in the left side, and it can't refuse all defect on right side.If when critical value was low, then the defective on right side will be still less, still, the erroneous decision in left side will be more, and vice versa.According to the present invention, partly can obtain the good efficacy of highly significant at low frequency, and, when when erroneous judgement takes place in radio-frequency head, can departure.
Fig. 4 a and 4b show the measurement according to embodiments of the invention and known median filter respectively.Clear as can be known by Fig. 4 a and 4b, when the amplitude of known median filter is subjected to heavy damage, still very strong according to embodiments of the invention, influenced hardly.
Though shown especially with reference to preferred embodiment and the present invention be described,, those of ordinary skills can do different changes and modification not deviating under scope of invention and the spirit.Therefore, scope of invention is only determined by accompanying Claim.
Main reference numerals list
200 dynamic defect means for correctings
The 202-208 delay element
210 median filters
220 maximums are found device
222 low pass filters
230 minimum values are found device
232 low pass filters
240 comparators
The 260-264 multiplexer
270 calculation elements
280 bed devices gradually
Claims (14)
1. dynamic defect correction method is used for the defective of correction pixels, comprises the steps:
Input step is imported the signal strength signal intensity of pending pixel and the signal strength signal intensity that this pending pixel reaches a plurality of pixels afterwards before;
Comparison step compares the signal strength signal intensity that reaches a plurality of pixels afterwards before this mutually, to find out maximum signal value and minimum signal strength value;
Average step, the signal strength signal intensity of nearest pixel is average after reaching recently before the pixel that this is pending, to obtain average intensity value;
Determining step, according to the signal strength signal intensity of this pending pixel respectively with the comparative result of this minimum signal strength value and this maximum signal value, judge whether this pending pixel is the tool defective pixels;
Aligning step when this pending pixel is judged as when having defective, determines a layer parameter k gradually according to the comparative result of obtaining in this determining step, to proofread and correct the signal strength signal intensity of this pending pixel.
2. the method for claim 1, wherein, in this determining step, when the signal strength signal intensity of this pending pixel is not less than this minimum signal strength value and is not more than this maximum signal value, this pending pixel is judged as does not have defective, and when the signal strength signal intensity of this pending pixel during less than this minimum signal strength value or greater than this maximum signal value, this pending pixel is judged as has defective.
3. the method for claim 1, wherein, when the signal strength signal intensity of this pending pixel during less than this minimum signal strength value, this pending pixel is judged as has defective, and, wherein, the difference between this average intensity value and this maximum signal value is passed through normalization to obtain this gradually layer parameter k, and obtain output valve with following formula, as the correction value output of the intensity of this pending signal:
(1-k) * (this average intensity value)+k* (this minimum signal strength value).
4. the method for claim 1, wherein, when the signal strength signal intensity of this pending pixel during greater than this maximum signal value, this pending pixel is judged as has defective, and, wherein, the difference between this average intensity value and this maximum signal value is passed through normalization to obtain this gradually layer parameter k, and obtain output valve with following formula, as the correction value output of the intensity of this pending signal:
(1-k) * (this average intensity value)+k* (this maximum signal value).
5. as the described method of one of claim 1 to 4, wherein, should before and after a plurality of pixels be two contiguous pixels before this pending pixel and two contiguous pixels afterwards.
6. dynamic defect means for correcting is used for the defective of correction pixels, comprising:
Input unit is in order to import the adjacent a plurality of contiguous pixels signals in pending picture element signal and front and back;
The maximum determination device, according to these adjacent a plurality of contiguous pixels signal strength signal intensities with decision maximum signal value;
The minimum value determination device, according to these adjacent a plurality of contiguous pixels signal strength signal intensities with decision minimum signal strength value;
In value determining device, according to will this pending pixel nearest before and the signal strength signal intensity of pixel afterwards, to obtain average intensity value;
Judgment means according to signal strength signal intensity, this maximum signal strength values, this minimum signal strength value of this pending pixel, determines whether the signal of the pixel that this is pending needs to proofread and correct;
The difference calculation element, output according to this judgment means, calculate the difference between this average intensity value and this maximum signal value or this minimum signal strength value, and, comprise the limit value device, when this difference during less than critical value, with this difference original state output, when this difference during greater than this critical value, this difference is set at equals preset value, the output of this limit value device is exported as parameter value k through normalization;
Bed device gradually, according to output, this average intensity value, this parameter value k of this judgment means, the value of calculating with following formula is as output signal:
Output valve=(1-k) * this average intensity value+k * I,
I depends on the output of this judgment means and is this large-signal intensity level or this minimum signal strength value,
Wherein, when determining the signal of the pixel that this is pending, this judgment means do not need timing, the signal of the pixel that this is pending can original state be exported, when this judgment means determines the signal demand timing of the pixel that this is pending, with this gradually the output signal of bed device be used as the signal output of this pending pixel.
7. device as claimed in claim 6, wherein, this critical value be this pending pixel signal strength signal intensity 1/20 to 1/5.
8. device as claimed in claim 6, wherein, when the signal strength signal intensity of this pending pixel was the numerical value of 0-255, this critical value was in 4 to 18 scope.
9. device as claimed in claim 6 further comprises first low pass filter and second low pass filter, respectively in order to the output signal of this maximum determination device of low-pass filtering and this minimum value determination device.
10. as the described device of one of claim 6 to 9, wherein, the adjacent a plurality of picture element signals in these front and back are continuous two picture element signals after continuous two picture element signals before this pending picture element signal reach.
11. an electrophotographic device uses as the described method of one of claim 1 to 4 and carries out treatment for correcting with the picture element flaw to imager.
12. an electrophotographic device comprises imager, and uses method as claimed in claim 5 to carry out treatment for correcting with the picture element flaw to this imager.
13. an electrophotographic device comprises as the described device of one of claim 6 to 8.
14. an electrophotographic device comprises device as claimed in claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100488656A CN100550979C (en) | 2005-12-31 | 2005-12-31 | Dynamic defect correction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100488656A CN100550979C (en) | 2005-12-31 | 2005-12-31 | Dynamic defect correction method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1992792A true CN1992792A (en) | 2007-07-04 |
| CN100550979C CN100550979C (en) | 2009-10-14 |
Family
ID=38214708
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100488656A Expired - Fee Related CN100550979C (en) | 2005-12-31 | 2005-12-31 | Dynamic defect correction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100550979C (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102550017A (en) * | 2009-10-05 | 2012-07-04 | 佳能株式会社 | Defect detecting method for imaging device, and imaging device |
-
2005
- 2005-12-31 CN CNB2005100488656A patent/CN100550979C/en not_active Expired - Fee Related
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102550017A (en) * | 2009-10-05 | 2012-07-04 | 佳能株式会社 | Defect detecting method for imaging device, and imaging device |
| CN102550017B (en) * | 2009-10-05 | 2014-12-10 | 佳能株式会社 | Defect detecting method for imaging device, and imaging device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100550979C (en) | 2009-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP4054184B2 (en) | Defective pixel correction device | |
| JP4127411B1 (en) | Image processing apparatus and method | |
| CN110473189B (en) | Text image definition judging method and system | |
| JP4668185B2 (en) | Image processing method | |
| CN1783946A (en) | Systems and methods for image processing and video player for realizing by the method | |
| US20080112639A1 (en) | Method and apparatus for removing noise in dark area of image | |
| US9959633B2 (en) | Texture detection in image processing | |
| US20060017824A1 (en) | Image processing device, image processing method, electronic camera, and scanner | |
| EP1394742B1 (en) | Method for filtering the noise of a digital image sequence | |
| CN1251146C (en) | Method and apparatus for reducing image noise | |
| JP4679174B2 (en) | Image processing apparatus and digital camera equipped with the image processing apparatus | |
| US7421143B2 (en) | Systems and methods for optimal dynamic range adjustment of scanned images | |
| CN1148777A (en) | Method for intensifying colour signal detail and circuit applied in colour video apparatus | |
| CN1992792A (en) | Dynamic defect correction method and apparatus, and electronic equipment thereof | |
| US7352397B2 (en) | Circuit and method for contour enhancement | |
| JP4166974B2 (en) | Pixel scratch detection / correction device | |
| US11544862B2 (en) | Image sensing device and operating method thereof | |
| CN115359022A (en) | Power chip quality detection method and system | |
| KR101043061B1 (en) | SMD test method using the discrete wavelet transform | |
| US11528426B2 (en) | Image sensing device and operating method thereof | |
| US20090180709A1 (en) | Providing a digital copy of a source image | |
| US20060192865A1 (en) | Noise processing apparatus and image pickup apparatus | |
| CN112954232A (en) | Moving image processing method, moving image processing apparatus, camera, and storage medium | |
| Stach et al. | Use of web cameras for watermark detection | |
| JP2006191231A (en) | Method for inspecting camera module |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20091014 Termination date: 20131231 |