CN102316283B - Real-time light leak phenomenon removing device for space and line space transfer CCD (charge coupled device) camera - Google Patents
Real-time light leak phenomenon removing device for space and line space transfer CCD (charge coupled device) camera Download PDFInfo
- Publication number
- CN102316283B CN102316283B CN 201110272780 CN201110272780A CN102316283B CN 102316283 B CN102316283 B CN 102316283B CN 201110272780 CN201110272780 CN 201110272780 CN 201110272780 A CN201110272780 A CN 201110272780A CN 102316283 B CN102316283 B CN 102316283B
- Authority
- CN
- China
- Prior art keywords
- light leak
- view data
- code value
- pixel
- row
- 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
- Transforming Light Signals Into Electric Signals (AREA)
Abstract
The invention relates to a real-time light leak phenomenon removing device for a space and line space transfer CCD (charge coupled device) camera, which comprises a device for collecting original image data sensed in effective pixel area of the CCD, a device for extracting the light leak signal numeral code in a dark pixel area of the CCD, and an image data correcting device, wherein the image data correcting device is used for correcting corresponding column data of the original image by utilizing the light leak signal numeral code extracted from the dark pixel area of the CCD, and the image data of which the light leak signal is removed can be obtained. According to the invention, the column data of the original image is corrected by utilizing the light leak signal numeral code extracted from the dark pixel area of the CCD, the calculated amount is small, the real time is good, and since the light leak signal can accurately stand for light leak charge of a high light area, the light leak phenomenon is better removed without changing the CCD working conditions.
Description
Technical field
The invention belongs to the space camera technical field, relate to a kind of real-time cancellation element of light leak of space camera.
Background technology
Along with the develop rapidly of aeronautical and space technology, big face battle array picture formula CCD camera is with its high-resolution, and the advantage of big visual field becomes the main developing direction of Aero-Space camera.In several area array CCD structures, the interline transfer area array CCD has no smear, transfer rate is fast, have characteristics such as electronic shutter function, be particularly suitable for target acquisition and identification, resource exploration, work such as geological mapping, but at transition surface array CCD duration of work in the ranks, can produce light leakage phenomena, image quality is impacted.
Light leak is a kind of noise that takes place when ccd image sensor transmission signal charge, also is the distinctive noise of Array CCD transducer.Occur light leak easily when the light that contains high brightness in the reference object or sunlight, the part upper and lower that shows as high brightness in image can be seen the noise of wire or banded white.Light leak appears at the above-below direction of high brightness reference object, and it forms reason and is: during the pixel of strong illumination in the photography zone light leak electric charge can take place, the light leak electric charge mixes with the signal charge that vertical CCD transmits.Because the useful signal electric charge shifts among the vertical CCD of the blank that finishes, residual the light leak electric charge of lasting generation, these light leak electric charges mix with subsequent frame useful signal electric charge, the result causes the interior light leak of successive frame to appear at the upper and lower in photography zone, thereby form wire or banded light leak noise, look like the lines of vertically pulling out.The main cause that interline transfer area array CCD light leak produces comprises:
1, from the light of photodiode opening incident, the incident angle composition vertical with the surface arranged not only.Because light has the character of fluctuation, has the ability of diffusion by the light of opening, even reading edge or opposition side near the grid, also can carry out a spot of opto-electronic conversion.The electronics of Fa Shenging herein enters vertical CCD along photodiode with Potential Distributing between the vertical CCD, becomes the light leak electric charge.
2, at the edge of opening of photodiode, namely the boundary of different refractivity such as Si substrate surface reflects, the light of scattering enters vertical CCD, and the electric charge of Chan Shenging also can cause light leak herein.
3, on the surface of burying the type photodiode, the electronics that the P+ zone takes place is in diffusion into the surface, and the electric charge that leaks past adjacent vertical CCD also can cause light leak.
4, the AL film defectiveness that carries out shading owing to vertical CCD, the light that bleeds also can cause light leak to take place.
Light leak inhibition method commonly used at present has:
One, regulates the electronic shutter signal sequential
The order of severity of light leakage phenomena depends on the time for exposure of CCD, and the time for exposure is more long, and the signal charge in high luminance targets zone is more many, and light leakage phenomena is more serious.By the radiation calibration experiment, the number value of the high luminance targets correspondence when the searching light leakage phenomena occurs is regulated electronic shutter signal as with reference to value, the number value of image output is remained below under the situation of reference value, thereby reach the purpose that suppresses light leakage phenomena.Concrete electronic shutter control method flow process is seen Fig. 1: the analog signal of CCD image device output is sent into image processing module after quantizing through the video AD chip; Image processing module calculates every two field picture gray average, and itself and reference value are compared, and provides feedback signal according to comparative result again; The feedback signal one tunnel of image processing module output is sent into the driving signaling control unit, and to change space camera electronic shutter quantity, the one tunnel sends into the video AD chip to adjust gain.
At the residing particular surroundings of space camera, image processing module provides feedback signal according to the comparative result of every two field picture gray average and reference value, control electronic shutter signal sequential, control CCD photosensitive unit exposure the time of integration to adjust, when the gradation of image average near reference value, and when continue adjusting the electronic shutter gear and will exceed reference value and cause the overshoot concussion, adjust by gain, make gray average satisfy term of reference.By gain control, issuable overshoot concussion in the time of can reducing the adjustment of electronic shutter gear.The method of regulating electronic shutter can effectively be alleviated light leakage phenomena, but can not rely on the electronic shutter method merely, crosses the low gradation of image that reduces, and can influence camera dynamic range, reduces signal to noise ratio.
Two, follow-up digital image processing method
In image, eliminate the influence of light leak signal fully, can only be by the method for follow-up Digital Image Processing, digital image processing method commonly used at present is by mainly being when rejecting false target after target detection, adopts the false target of removing in the target hangover to realize.Step 1: for each row of input picture F, calculate the average gray of every row image.Step 2: for each pixel of image F, by formula 1 calculate the gradation of image matrix g that eliminates behind the light leak (i, j).
The shortcoming of this method is that very high for target area number value, smear zone or background area are digital to be worth when hanging down, and adopts mean value to replace the light leak signal, can lose a part of useful half-tone information.
Though above-mentioned two kinds of methods can play a role to the inhibition of light leakage phenomena, all can not fundamentally eliminate the peculiar noise of this CCD device.And for the Aero-Space area array CCD camera, after camera orbit altitude and camera flight attitude are fixing, we can't be random change CCD pixel readout clock, can not arbitrarily change the CCD time for exposure, moreover reduce the method for time for exposure and can influence the CCD image quality.Though can eliminate light leakage phenomena by the method that successive image is handled, for big area array CCD, successive image is handled need open up a large amount of memory spaces, and amount of calculation is big, has influenced the real-time of imaging system.
Summary of the invention
The technical problem to be solved in the present invention provide a kind of can be under the situation that does not change the CCD condition of work, eliminate the real-time cancellation element of transition surface array CCD camera light leakage phenomena between the spatial row of light leakage phenomena well.
In order to solve the problems of the technologies described above, the real-time cancellation element of transition surface array CCD camera light leakage phenomena comprises between spatial row of the present invention:
Gather the device of the raw image data of area array CCD effective pixel area induction;
Extract the device of the dark pixel of area array CCD zone light leak number of signals code value;
View data means for correcting: utilize and from the light leak number of signals code value of the dark pixel extracted region of area array CCD original image respective column data to be proofreaied and correct, obtain removing the view data behind the light leak signal.
The present invention forms by analyzing the CCD structure, according to comprising these characteristics of light leak signal in the dark pixel zone, utilizes and from the light leak signal of dark pixel extracted region the raw image data respective column is proofreaied and correct, and amount of calculation is little, and real-time is good.And because the light leak signal can represent the light leak electric charge that high-brightness region produces exactly, do not change under the situation of CCD condition of work in assurance, eliminated light leakage phenomena well.
The device of the dark pixel of described extraction area array CCD zone light leak number of signals code value comprises a filter, described filter carries out filtering to the light leak signal from the dark pixel extracted region of area array CCD, eliminate noise, then filtered result is delivered to the view data means for correcting, utilize the light leak number of signals code value that comprises among the filtered result that original image respective column data are proofreaied and correct by the view data means for correcting.
Described filter adopts median filter.
Because there is the black level noise in dark pixel zone, directly the light leak signal is removed from the pixel count code value (being gray value) of original image respective column, can produce coarse elimination result, it shows as the degeneration that produces vision in image, such as some additional wire noises.In order to eliminate these noise effects, the present invention carries out filtering with the light leak signal that extracts, and remains with the light leak signal of usefulness, eliminates noise.Usually filter is divided into linear filter and nonlinear filter two big classes.Linear filter has smoothing effect preferably to Gaussian noise, but other Noise Suppression effect is relatively poor, and can fuzzy edge.Median filter in the nonlinear filter can well be protected the edge profile information in filtered noise, it is very useful, particularly particularly effective for binary noise to the interference of eliminating isolated point and line segment.This point meets the filtering requirements of light leak signal especially, handles so adopt median filter that image is carried out filtering among the present invention.
Described median filter is the Fast Median Filtering device.
Described Fast Median Filtering device comprises first memory block, second memory block, frequency divider, sorting unit, the first register group, the second register group; As writing clock, storage is from the capable view data of N of the dark pixel extracted region of area array CCD with the image pixel clock in described first memory block; Frequency divider with the image pixel clock multiplier after, as the clock that reads of first memory block; Sorting unit reads view data in first memory block by the principle of same column, and sorting by the ascending order of same column pixel count code value generates the capable view data of new N, and the capable view data of new N writes second memory block; The view data of N row before the first register group reads from second memory block, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the first register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; The second register group reads the view data of secondary series to the N+1 row from second memory block simultaneously, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the second register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; After sorting unit outputs to the view data means for correcting with median, the first register group and the second register group zero clearing; Then, second memory block is pressed ping-pong structure and is given the first register group with the 3rd row to N+2 row view data assignment, be listed as to the N+3 row view data assignment second register group the 4th, continued the pixel count code value in the first register group and the second register group is carried out line ordering and diagonal ordering taking-up median by sorting unit, median is outputed to the view data means for correcting; So analogize, until the intermediate value ordering of finishing the capable view data of whole N and median output; The view data means for correcting utilizes the light leak number of signals code value that comprises in the median that original image respective column data are proofreaied and correct, and obtains removing the view data behind the light leak signal.
In order to satisfy the system real time requirement, finishing hardware handles in real time, need before arriving, the effective pixel data of CCD finish medium filtering, and traditional median filtering algorithm ordering is time-consuming, be unfavorable for real-time processing, the present invention utilizes the improvement algorithm of grey scale pixel value ordering in the window, it is the Fast Median Filtering algorithm, obtained comprising the median ordered series of numbers of light leak signal, and utilize the median ordered series of numbers that raw image data is proofreaied and correct by the view data means for correcting, in contrast to traditional median filtering algorithm, not only operand significantly reduces, and the hardware resource that takies also still less, so it is faster than traditional median filtering algorithm to the processing speed of view data.
Described view data means for correcting comprises a subtracter; Subtracter with light leak number of signals code value from the digital value of the original image pixels of light leak signal same column deduct, obtain removing the view data behind the light leak signal.
Subtracter utilizes the view data after formula (2) obtains removing the light leak signal;
I
y(i,j)=I
x(i,j)-y
p(j)
(2)
I in the formula (2)
y(i is that capable, the j row pixel of i is removed number value behind the light leak signal, I on the area array CCD effective pixel area j)
x(i j) is the digital value of i original image capable, j row pixel on the area array CCD, y
p(j) the j row pixel light leak number of signals code value for extracting.
Described view data means for correcting can also comprise comparator and weighted average arithmetic element; Comparator compares light leak number of signals code value and saturation threshold, if light leak number of signals code value is less than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct, obtain removing the view data behind the light leak signal; If light leak number of signals code value is greater than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct after, be weighted view data after on average obtaining proofreading and correct by regional pixel count code value about weighted average arithmetic element utilization generation light leakage phenomena zone again.
If light leak number of signals code value is greater than saturation threshold, n row pixel is occupied in the zone that light leakage phenomena takes place,
Then the weighted average arithmetic element is utilized the view data after formula (3) obtains removing the light leak signal;
(3)
I wherein
y(i, j), I
y(i, j+1) ..., I
y(i, j+n-1) the number value that obtains after calibrated for each row pixel of zone of saturation, I
y(i, j-n), I
y(i, j-n-1) ..., I
y(i, j-1) the number value that obtains by zone of saturation left side after calibrated of the n row pixel of getting, I
y(i, j+n), I
y(i, j+n+1) ..., I
y(i, j+2n-1) for the right side, zone of saturation get number value behind the n row pixel correction,
For getting the weights of each row pixel on the left of the zone of saturation and right side;
Can get 1 * (1/n) numerical value that obtains after rounding,
Can go bail for and stay the numerical value of 1 decimal.
Described view data means for correcting can also adopt following technical proposals:
The view data means for correcting comprises subtracter, comparator and weighted average arithmetic element; Comparator compares light leak number of signals code value and saturation threshold, if light leak number of signals code value is less than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct, obtain removing the view data behind the light leak signal; If light leak number of signals code value is greater than saturation threshold, then be weighted view data after on average obtaining proofreading and correct by regional pixel count code value about weighted average arithmetic element utilization generation light leakage phenomena zone.
If light leak number of signals code value is greater than saturation threshold, n row pixel is occupied in the zone that light leakage phenomena takes place, and then the weighted average arithmetic element is utilized the view data after formula (4) obtains removing the light leak signal;
(4)
I wherein
y(i, j), I
y(i, j+1) ..., I
y(i, j+n-1) the number value that obtains after calibrated for each row pixel of zone of saturation, I
x(i, j-n), I
x(i, j-n-1) ..., I
x(i j-1) is got the digital value of n row pixel original image, I by the left side, zone of saturation
x(i, j+n), I
x(i, j+n+1) ..., I
x(i, j+2n-1) by the right side, zone of saturation got the digital value of n row pixel original image,
For getting the weights of each row pixel on the left of the zone of saturation and right side;
Can go bail for and stay the numerical value of 1 decimal.
In fact, because the Fast Median Filtering device is from the capable view data of the dark pixel extracted region of area array CCD N, have only the row pixel count code value corresponding with the light leak zone non-vanishing, and other row pixel count code values are zero, also have only the row median corresponding with the light leak zone non-vanishing in the median sequence that obtains.Thereby I
y(i, j-n), I
y(i, j-n-1) ..., I
y(i, j-1), I
y(i, j+n), I
y(i, j+n+1) ..., I
y(i j+2n-1) equals I respectively
x(i, j-n), I
x(i, j-n-1) ..., I
x(i, j-1), I
x(i, j+n), I
x(i, j+n+1) ..., I
x(i, j+2n-1).Wherein the selection of N does not have strict restriction, can require to select according to the dark pixel line number of CCD with to real-time.
If the digital value of actual effective pixel area does not have saturated, the area image gray value that light leakage phenomena takes place is the result of true picture gray value and the stack of light leak signal gray value, with light leak signal gray value from the original image pixels gray value of light leak signal same column deduct and can reduce true picture accurately.But after the digital value generation of actual effective pixel area is saturated, the light leak signal is directly deducted from view data, can bring image degradation to raw image data, simple understanding is exactly that the output image of removing the light leak signal can not keep the raw image data details fully.Whether saturatedly from light leak number of signals code value, can judge with the digital value of the actual effective pixel area of light leak signal same column.At this moment, by setting up saturation threshold a can judge whether have saturation signal in the valid pixel view data, if the valid pixel view data saturated then can not be simple remove the light leak signal, need compensate.When the light leak electric charge of high-brightness region generation is excessive, the light leak zone greater than saturation threshold is weighted average compensation, the whole CCD camera can be operated in the very big dynamic range, finally eliminate light leakage phenomena.
Description of drawings
Below in conjunction with the drawings and specific embodiments the present invention is described in further detail.
Fig. 1 is electronic shutter light-dimming method theory diagram.
Fig. 2 is the structured flowchart of the real-time cancellation element of transition surface array CCD camera light leakage phenomena between spatial row of the present invention.
Fig. 3 is Fast Median Filtering device structured flowchart.
Fig. 4 carries out the schematic diagram of sequencer procedure to 3 row, 3 columns certificate for sorting unit.
Fig. 5 is the medium filtering flow chart.
Fig. 6 is view data means for correcting structured flowchart among the embodiment 2.
View data means for correcting structured flowchart among Fig. 7 embodiment 3.
Fig. 8 weighted average compensation principle schematic diagram.
Fig. 9 a light leakage phenomena original image.
Fig. 9 b, 9c light leak result picture.
Embodiment
The real-time cancellation element of transition surface array CCD camera light leakage phenomena utilizes the FPGA programming to realize between spatial row of the present invention.
As shown in Figure 2, described FPGA inside comprises the following function module:
Gather the device of the raw image data of area array CCD effective pixel area induction;
Extract the device of the dark pixel of area array CCD zone light leak number of signals code value;
View data means for correcting: utilize and from the light leak number of signals code value of the dark pixel extracted region of area array CCD original image respective column data to be proofreaied and correct, obtain removing the view data behind the light leak signal.
As shown in Figure 2, the real-time cancellation element of transition surface array CCD camera light leakage phenomena comprises between spatial row of the present invention:
Gather the device of the raw image data of area array CCD effective pixel area induction;
Extract the device of the dark pixel of area array CCD zone light leak number of signals code value;
View data means for correcting: utilize and from the light leak number of signals code value of the dark pixel extracted region of area array CCD original image respective column data to be proofreaied and correct, obtain removing the view data behind the light leak signal.
The device of the dark pixel of described extraction area array CCD zone light leak number of signals code value: extract light leak number of signals code value in the one-row pixels arbitrarily by the dark pixel of area array CCD zone, and give the view data means for correcting with light leak number of signals code value.
Described view data means for correcting comprises a subtracter, and this subtracter utilizes formula (2) that light leak number of signals code value is deducted from the respective column raw image data, obtains removing the view data behind the light leak signal;
I
y(i,j)=I
x(i,j)-y
p(j) (2)
I in the formula (2)
y(i is that capable, the j row pixel of i is removed number value behind the light leak signal, I on the area array CCD effective pixel area j)
x(i j) is the digital value of i original image capable, j row pixel on the area array CCD, y
p(j) be the light leak number of signals code value of the j row pixel of extraction.
As shown in Figure 2, the real-time cancellation element of transition surface array CCD camera light leakage phenomena comprises between spatial row of the present invention:
Gather the device of the raw image data of area array CCD effective pixel area induction;
Extract the device of the dark pixel of area array CCD zone light leak number of signals code value;
View data means for correcting: utilize and from the light leak number of signals code value of the dark pixel extracted region of area array CCD original image respective column data to be proofreaied and correct, obtain removing the view data behind the light leak signal.
The device of the dark pixel of described extraction area array CCD zone light leak number of signals code value comprises a filter, and this filter adopts the Fast Median Filtering device.
Shown in Fig. 3,4, described Fast Median Filtering device comprises first memory block, second memory block, frequency divider, sorting unit, the first register group, the second register group; As writing clock, storage is from 3 row view data of the dark pixel extracted region of area array CCD with the image pixel clock in described first memory block; Frequency divider with image pixel clock 6 frequencys multiplication after, as the clock that reads of first memory block; Sorting unit reads view data in first memory block by the principle of same column, and sorting by the ascending order of same column pixel count code value generates 3 new row view data, and 3 new row view data write second memory block; The first register group reads the view data of first three columns from second memory block, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the first register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; Simultaneously the second register group from second memory block, read second and third, the view data of four row, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the second register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; After sorting unit outputs to the view data means for correcting with median, the first register group and the second register group zero clearing; Then, second memory block press ping-pong structure with third and fourth, five row view data assignment give the first register group, with fourth, fifth, six row view data assignment, the second register group, continued the pixel count code value in the first register group and the second register group is carried out line ordering and diagonal ordering taking-up median by sorting unit, median is outputed to the view data means for correcting; So analogize, until the intermediate value ordering of finishing whole 3 row view data and median output; The view data means for correcting utilizes the light leak number of signals code value that comprises in the median that original image respective column data are proofreaied and correct, and obtains removing the view data behind the light leak signal.
Whether saturatedly from light leak number of signals code value, can judge with the number value of the actual effective pixel area respective column pixel of light leak signal same column, can judge by setting up saturation threshold a this moment, whether there is saturation signal in the valid pixel view data, if the valid pixel view data saturated then can not be simple remove the light leak signal, need compensate.
By radiation calibration experiment can draw when light leak regional signal amplitude when 200 digital values are above, actual effective pixel area respective column pixel count code value just takes place saturated, therefore threshold value a can be set in 200 digital values, when needs compensate, adopt weighted-average method, utilize that regional pixel value is weighted on average about the zone of saturation.
As shown in Figure 6, described view data means for correcting comprises subtracter, comparator and weighted average arithmetic element; Comparator compares light leak number of signals code value and saturation threshold, if light leak number of signals code value is less than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct, obtain removing the view data behind the light leak signal; If light leak number of signals code value is greater than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct after, be weighted view data after on average obtaining proofreading and correct by regional pixel count code value about the zone of weighted average arithmetic element utilization generation light leakage phenomena.
Subtracter utilizes formula (2) that light leak number of signals code value is deducted from the respective column raw image data, obtains removing the view data behind the light leak signal;
I
y(i,j)=I
x(i,j)-y
p(j) (2)
I in the formula (2)
y(i is that capable, the j row pixel of i is removed number value behind the light leak signal, I on the area array CCD effective pixel area j)
x(i j) is the digital value of i original image capable, j row pixel on the area array CCD, y
p(j) be the light leak number of signals code value of the j row pixel of extraction.
If light leak number of signals code value, is supposed the zone that light leakage phenomena takes place less than saturation threshold and is occupied 3 row pixels, utilizes the view data after formula (5) calculates correction.
(5)
I wherein
y(i, j), I
y(i, j+1), I
y(i, j+2), j, j+1 capable for the i that light leakage phenomena takes place, j+2 row pixel are removed the number value behind the light leak signal;
Be the weights of j-3, j-2, j-1, j+3, j+4, j+5 row; I
y(i, j-3), I
y(i, j-2), I
y(i, j-1), I
y(i, j+3), I
y(i, j+4), I
y(i j+5) is respectively original image j-3, j-2, j-1, j+3j+4, j+5 row pixel count code value.
As shown in Figure 2, the real-time cancellation element of transition surface array CCD camera light leakage phenomena comprises between spatial row of the present invention:
Gather the device of the raw image data of area array CCD effective pixel area induction;
Extract the device of the dark pixel of area array CCD zone light leak number of signals code value;
View data means for correcting: utilize and from the light leak number of signals code value of the dark pixel extracted region of area array CCD original image respective column data to be proofreaied and correct, obtain removing the view data behind the light leak signal.
The described filter that comprises of device of the dark pixel of described extraction area array CCD zone light leak number of signals code value, this filter adopts the Fast Median Filtering device.
As shown in Figure 3, described Fast Median Filtering device comprises first memory block, second memory block, frequency divider, sorting unit, the first register group, the second register group; As writing clock, storage is from 5 row view data of the dark pixel extracted region of area array CCD with the image pixel clock in described first memory block; Frequency divider with image pixel clock 12 frequencys multiplication after, as the clock that reads of first memory block; Sorting unit reads view data in first memory block by the principle of same column, and sorting by the ascending order of same column pixel count code value generates 5 new row view data, and 5 new row view data write second memory block; The first register group reads the view data of the first five row from second memory block, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the first register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; The second register group reads the view data of the second~six row from second memory block simultaneously, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the second register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; After sorting unit outputs to the view data means for correcting with median as light leak number of signals code value, the first register group and the second register group zero clearing; Then, second memory block is pressed ping-pong structure and is given the first register group with the three~seven row view data assignment, with the four~eight row view data assignment second register group, continued the pixel count code value in the first register group and the second register group is sorted and diagonal ordering taking-up median by sorting unit, median is outputed to the view data means for correcting as light leak number of signals code value; So analogize, until the intermediate value ordering of finishing whole 5 row view data and median output; The view data means for correcting utilizes the light leak number of signals code value that comprises in the median that original image respective column data are proofreaied and correct, and obtains removing the view data behind the light leak signal.
Whether saturatedly from light leak number of signals code value, can judge with the number value of the actual effective pixel area respective column pixel of light leak signal same column, can judge by setting up saturation threshold a this moment, whether there is saturation signal in the valid pixel view data, if the valid pixel view data saturated then can not be simple remove the light leak signal, need compensate.
By radiation calibration experiment can draw when light leak regional signal amplitude when 200 digital values are above, actual effective pixel area respective column pixel count code value just takes place saturated, therefore threshold value a can be set in 200 digital values, when needs compensate, adopt weighted-average method, utilize that regional pixel value is weighted on average about the zone of saturation.
As shown in Figure 7, described view data means for correcting comprises subtracter, comparator and weighted average arithmetic element; Comparator compares light leak number of signals code value and saturation threshold, if light leak number of signals code value is less than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct, obtain removing the view data behind the light leak signal; If light leak number of signals code value is greater than saturation threshold, then be weighted view data after on average obtaining proofreading and correct by regional pixel count code value about the zone of weighted average arithmetic element utilization generation light leakage phenomena.
If light leak number of signals code value is less than saturation threshold, then subtracter utilizes formula (2) that light leak number of signals code value is deducted from the respective column raw image data, obtains removing the view data behind the light leak signal;
I
y(i,j)=I
x(i,j)-y
p(j) (2)
I in the formula (2)
y(i is that capable, the j row pixel of i is removed number value behind the light leak signal, I on the area array CCD effective pixel area j)
x(i j) is the digital value of i original image capable, j row pixel on the area array CCD, y
p(j) be the light leak number of signals code value of the j row pixel of extraction.
If light leak number of signals code value, is supposed the zone that light leakage phenomena takes place less than saturation threshold and is occupied 3 row pixels, utilizes the view data after formula (6) calculates correction.
(6)
I wherein
y(i, j), I
y(i, j+1), I
y(i, j+2), j, j+1 capable for the i that light leakage phenomena takes place, j+2 row pixel are removed the number value behind the light leak signal;
Weights for j-3, j-2, j-1, j+3, j+4, j+5 row; I
x(i, j-3), I
x(i, Ix-2), I
x(i, j-1), I
x(i, j+3), I
x(i, j+4), I
x(i j+5) is original image j-3, j-2, j-1, j+3j+4, j+5 row pixel count code value.
One, light leak signal extraction and filtering are handled
In order to satisfy the system real time requirement, finishing hardware handles in real time, need before arriving, the effective pixel data of CCD finish medium filtering, and the ordering of traditional median filtering algorithm time-consuming, be unfavorable for real-time processing, the present invention utilizes the improvement algorithm of grey scale pixel value ordering in a kind of window, i.e. Fast Median Filtering algorithm.The Fast Median Filtering algorithm in contrast to traditional median filtering algorithm, and not only operand significantly reduces, on FPGA hardware is realized shared hardware resource also still less, so it is faster than traditional median filtering algorithm to the processing speed of view data.Fig. 4 is the Fast Median Filtering schematic diagram.Ordered steps is as follows:
(1) a kind of ordering device of design, function is that 3 10 signless integers of A, B, C of input are put into the relevant position, ascending arrangement is opened;
(2) 3 column vectors of template data are sorted simultaneously, carry out water operation, generate a new matrix template;
(3) continue to use the ordering device that new matrix template is rearranged combination, second new matrix template of regeneration according to the row vector;
(4) the matrix leading diagonal vector of up-to-date generation is extracted, use the ordering device to sort again, take out intermediate value at last, this intermediate value is exactly the result that medium filtering will obtain.
That designs among the present invention realizes based on FPGA the key point of medium filtering is how to finish three numbers and reads simultaneously, and will utilize high-frequency clock clock as a comparison in the clock design, guarantees to draw the intermediate value result before the effective view data of CCD arrives.
Open up three ram in slice in FPGA inside as spatial cache A (being first memory block among the embodiment 2,3), first three rows data in the dark pixel zone are write in three ram in slice line by line.Adopt comparison clock that three numbers of same column are read in three number sorting units then and sort, by ascending order data are write after finishing and newly open up three ram in slice spatial cache B (being second memory block among the embodiment 2,3).Because three number ordering comparators based on the VHDL language design, finish three number orderings and need 4 clock cycle at most, finish 6 clock cycle of common needs so calculate above-mentioned action, the comparator clock adopts 6 times of pixel readout clock, guaranteed to finish in the time in one-row pixels the comparison of same column data, therefore finish the row ordering, needed for 1 row pixel readout time.
Because the ram in slice data are read and are called over, 3 data of line direction can't be read simultaneously, need open up the register group of 3*3 as the buffer register group of line ordering this moment.Three line data in the spatial cache B are read in the register group by 3*3 square template form, at first carry out line ordering, finish and carry out the diagonal ordering behind the line ordering, draw the intermediate value result after, with the zero clearing of whole register group.Above-mentioned action is finished to amount to needs 12 clock cycle.Because comparison clock is 6 frequencys multiplication of pixel readout clock, therefore need two register groups (being the first register group and the second register group among the embodiment 2,3) to carry out ping-pong operation, just can satisfy in 2 row pixels in readout time, finish the ordering of line ordering and diagonal, thereby draw the intermediate value result.So far whole filter design is finished substantially, and summing up needs the pixel of triplex row readout time.
Light leak signal extraction is characterised in that:
The extraction prerequisite of light leak signal is the effective imaging electric charge of dark pixel zone isolation, only stays light leak electric charge and other noise charge.The light leak electric charge shifts to horizontal transfer CCD with vertical transitions CCD under the effect of driving signal.In the effective charge zone, the light leak electric charge mixes with effective imaging electric charge.In dark pixel zone, pixel does not receive the ambient light photograph, and effectively the imaging electric charge is isolated, and only leaves the light leak electric charge and exists.Dark pixel area area image data are analyzed, and the dark pixel view data of minimum extraction 3 row is as the initial data of analyzing the light leak signal.
Real-time medium filtering based on FPGA is characterised in that:
If the capable pixel number of CCD is a, then open up the memory space A of a * 3 in FPGA inside, pixel clock as writing clock, is write memory space A with the ccd image data.
With pixel clock 6 frequencys multiplication, as the clock that reads of data in the memory space A.By same column principle reads image data, the line ordering of going forward side by side.Press new a of the ascending generation of ranking results * 3 pixel data matrixes, and it is written in the memory space B of a * 3 of opening up FPGA inside.
Set two 3 * 3 register group a
1, a
2A
9, b
1, b
2B
9, the view data in the memory space B is called over two group of 3 * 3 data, give a with its assignment
1, a
2A
9, b
1, b
2B
9Two register groups.At first the line data in the register group is sorted, and press ranking results and upgrade the register group, then the register group after upgrading is carried out the diagonal ordering, draw median.
During to register group assignment, still adopt 6 frequency doubling clocks of pixel readout clock, after data are finished ordering in the two register groups, namely to its zero clearing, press ping-pong structure and give two register groups with successive image data assignment, up to the intermediate value ordering of finishing whole data.
Two, the light leak signal is removed
We are y with filtered light leak signal definition
p(i), according to the principle of same column the light leak signal is unified to deduct from effective view data, the mathematical description of this process is:
I
y(i,j)=I
x(i,j)-y
p(j)
(2)
I in the formula (2)
y(i, j) and I
x(i, j) output image data and original input image data behind the representative removal light leak.Remove in the algorithm at the light leak signal, if the digital value of actual effective pixel area does not have saturated, the area image that light leakage phenomena takes place is true picture and the stack of light leak signal, by formula (1) view data of calculating back output can be reduced true picture accurately, but after the digital value generation of actual effective pixel area is saturated, the light leak signal is directly deducted from view data, can bring image degradation to raw image data, simple understanding is exactly that the output image of removing the light leak signal can not keep the raw image data details fully.Whether saturatedly from light leak number of signals code value, can judge with the digital value of the actual effective pixel area of light leak signal same column, can judge by setting up saturation threshold a this moment, whether there is saturation signal in the valid pixel view data, if the valid pixel view data saturated then can not be simple remove the light leak signal, need compensate.
By radiation calibration experiment draw when light leak regional signal amplitude when 200 digital values are above, actual effective pixel area view data just takes place saturated, therefore threshold value a is set in 200 digital values, when needs compensate, adopt weighted-average method, utilize that regional pixel value is weighted on average about the zone of saturation.
I
y(i,j)=(w
lgI
y(i,j′)+w
rgI
y(i,j″))/(w
l+w
r) (6)
The weighting compensation method is characterised in that in the light leak elimination algorithm:
Camera is carried out the radiation calibration experiment, adopt infinite distant place point-source of light as the object of observation, acquisition of image data.
Adjust point-source of light brightness, perspective image data, the digital value of document image, when view data mid point source region was saturated, record light leak number of signals code value was as saturation threshold.
When light leak number of signals code value during greater than saturation threshold, the view data row that deduct the light leak signal are weighted compensation.By radiation calibration experiment draw when light leak regional signal amplitude when 200 digital values are above, actual effective pixel area view data just takes place saturated, therefore threshold value a is set in 200 digital values, when needs compensate, adopt weighted-average method, utilize that regional pixel value is weighted on average about the zone of saturation.Process as shown in Figure 8.
I
y(i,j)=(w
lgI
y(i,j′)+w
rgI
y(i,j″))/(w
l+w
r)
The present invention is as follows to the removing method of light leakage phenomena:
The first, obtain the light leak signal.By analyzing the area array CCD structure, and electric charge branch mode, utilize the existing dark pixel of CCD zone, or utilize shield assembly manually to make some dark pixels zone, extract the light leak signal, the appearance of light leak is to form at high brightness reference object above-below direction, form cause the above-below direction influence reason as shown in Figure 1.Suppose in the regional Centromedian pixel of photography the light leak electric charge can take place when strong illumination, when the light leak electric charge was transferred to vertical transitions CCD from photosensitive area, the light leak signal can be formed on the straight line of top, photography zone in the initial picture frame.Because signal charge shifts among the vertical CCD of the blank that finishes, residual the light leak electric charge of lasting generation, therefore reading in the sequential next time, the signal charge that is positioned at photography middle section below also will add the light leak electric charge, the result causes the interior light leak of successive frame to appear at the upper and lower in photography zone, thereby forms wire or banded light leak noise.That is to say that light leak occurs from top to bottom equably, this is because vertical CCD constantly continues to shift with certain speed.Thereby as seen, the light leak signal in the dark pixel zone is the light leak signal that high luminance area produces in the imaging region, extracts the light leakage phenomena that the light leak signal gets final product the removal of images zone in dark pixel zone.
The second, utilize FPGA to be the core processing device, the light leak signal that extracts is carried out medium filtering to be handled, because there is the black level noise in dark pixel zone, simultaneously when brightness has surpassed availability of dynamic range, directly the light leak signal is removed from other valid pixels, can be produced coarse elimination result.It shows as the degeneration that produces vision in image, such as some additional wire noises.
In order to eliminate these noise effects, the light leak signal that we at first need to extract carries out filtering, remains with the light leak signal of usefulness, eliminates noise.Usually filter is divided into linear filter and nonlinear filter two big classes.Linear filter has smoothing effect preferably to Gaussian noise, but other Noise Suppression effect is relatively poor, and can fuzzy edge.Median filter in the nonlinear filter can also well be protected the edge profile information in filtered noise.It is very useful, particularly particularly effective for binary noise to the interference of eliminating isolated point and line segment.This point meets the filtering requirements of light leak signal especially, handles so adopted median filter that image is carried out filtering among the present invention.
Three, light leak is eliminated compensation
When target area brightness was too high, the light leak signal was near saturation value or be far longer than the number value of background area, directly deducts the light leak signal and can produce vision degenerate in the light leak signal area from original image, had namely lost the useful background image information of part.Therefore, need the zone that deduct high brightness light leak signal be compensated.
Good effect of the present invention:
The present invention is from dark pixel extracted region light leak signal, the mode that shifts of CCD electric charge by analysis, compare light leak removing method (get column mean and represent the light leak electric charge) in the past, can represent the light leak electric charge of high-brightness region generation from the light leak signal of dark pixel extracted region exactly.Utilize FPGA as the core processing device, designed the median filtering algorithm that is applicable to the light leak signal, the system delay of finishing filtering only was 3 row view data readout times, satisfy the requirement of handling in real time, by the radiation calibration experiment, determined saturation threshold, when the light leak electric charge of high-brightness region generation is excessive, light leak zone greater than saturation threshold is weighted average compensation, the whole CCD camera can be operated in the very big dynamic range, finally eliminate light leakage phenomena.The light leak removing method of this paper design adopts the hardware real-time processing method, and system real time is good, is applicable to most of interline transfer area array CCDs.
The present invention relates to the real-time method of eliminating of space CCD camera light leak signal.Utilize the method for dark pixel extracted region light leak signal, adopt FPGA as the core processing device, hardware is eliminated light leakage phenomena in real time.The present invention compares the condition of work that need not change space CCD camera with background technology, utilize the light leak signal of dark pixel extracted region, can reflect the light leak electric charge that high-brightness region produces exactly.Utilize FPGA as the core processing device, stronger operability is arranged, utilize FPGA to finish the real-time medium filtering of hardware, eliminate the black level noise in dark pixel zone, greatly improved the accuracy of light leak signal.Determine saturation threshold by radiation calibration experiment, with it as the reference value that is weighted average compensation.The present invention calculates simply, reliability is high, real-time good.The real-time removing method of light leak signal of the space CCD camera that the present invention proposes under the space environment, the application of the CCD camera that is restricted of condition of work has realistic meaning.
Claims (6)
1. the real-time cancellation element of transition surface array CCD camera light leakage phenomena between a spatial row is characterized in that comprising:
Gather the device of the raw image data of area array CCD effective pixel area induction;
Extract the device of the dark pixel of area array CCD zone light leak number of signals code value;
View data means for correcting: utilize and from the light leak number of signals code value of the dark pixel extracted region of area array CCD original image respective column data to be proofreaied and correct, obtain removing the view data behind the light leak signal;
The device of the dark pixel of described extraction area array CCD zone light leak number of signals code value comprises a filter, described filter carries out filtering to the light leak signal from the dark pixel extracted region of area array CCD, eliminate noise, then filtered result is delivered to the view data means for correcting, utilize the light leak number of signals code value that comprises among the filtered result that original image respective column data are proofreaied and correct by the view data means for correcting;
Described filter adopts the Fast Median Filtering device; Described Fast Median Filtering device comprises first memory block, second memory block, frequency divider, sorting unit, the first register group, the second register group; As writing clock, storage is from the capable view data of N of the dark pixel extracted region of area array CCD with the image pixel clock in described first memory block; Frequency divider with the image pixel clock multiplier after, as the clock that reads of first memory block; Sorting unit reads view data in first memory block by the principle of same column, and sorting by the ascending order of same column pixel count code value generates the capable view data of new N, and the capable view data of new N writes second memory block; The view data of N row before the first register group reads from second memory block, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the first register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; The second register group reads the view data of secondary series to the N+1 row from second memory block simultaneously, and sorted by the ascending order of pixel count code value of going together mutually by sorting unit, result by ordering upgrades the second register group, then sorting unit sorts by the ascending order of pixel count code value on the diagonal, takes out median; After sorting unit outputs to the view data means for correcting with median, the first register group and the second register group zero clearing; Then, second memory block is pressed ping-pong structure and is given the first register group with the 3rd row to N+2 row view data assignment, be listed as to the N+3 row view data assignment second register group the 4th, continued the pixel count code value in the first register group and the second register group is carried out line ordering and diagonal ordering taking-up median by sorting unit, median is outputed to the view data means for correcting; So analogize, until the intermediate value ordering of finishing the capable view data of whole N and median output; The view data means for correcting utilizes the light leak number of signals code value that comprises in the median that original image respective column data are proofreaied and correct, and obtains removing the view data behind the light leak signal.
2. the real-time cancellation element of transition surface array CCD camera light leakage phenomena between spatial row according to claim 1 is characterized in that described view data means for correcting comprises a subtracter; Subtracter with light leak number of signals code value from the digital value of the original image pixels of light leak signal same column deduct, obtain removing the view data behind the light leak signal.
3. the real-time cancellation element of transition surface array CCD camera light leakage phenomena between spatial row according to claim 2 is characterized in that described view data means for correcting also comprises comparator and weighted average arithmetic element; Comparator compares light leak number of signals code value and saturation threshold, if light leak number of signals code value is less than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct, obtain removing the view data behind the light leak signal; If light leak number of signals code value is greater than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct after, be weighted view data after on average obtaining proofreading and correct by regional pixel count code value about weighted average arithmetic element utilization generation light leakage phenomena zone again.
4. the real-time cancellation element of transition surface array CCD camera light leakage phenomena between spatial row according to claim 3 is characterized in that described weighted average arithmetic element utilizes the view data after formula (3) obtains removing the light leak signal;
Wherein, the pixel columns that n occupies for the zone that light leakage phenomena takes place, I
y(i, j), I
y(i, j+1) ..., I
y(i, j+n-1) the number value that obtains after calibrated for each row pixel of zone of saturation, I
y(i, j-n), I
y(i, j-n-1) ..., I
y(i, j-1) the number value that obtains by zone of saturation left side after calibrated of the n row pixel of getting, I
y(i, j+n), I
y(i, j+n+1) ..., I
y(i, j+2n-1) for the right side, zone of saturation get number value behind the n row pixel correction,
For getting the weights of each row pixel on the left of the zone of saturation and right side;
5. the real-time cancellation element of transition surface array CCD camera light leakage phenomena between spatial row according to claim 1 is characterized in that the view data means for correcting comprises subtracter, comparator and weighted average arithmetic element; Comparator compares light leak number of signals code value and saturation threshold, if light leak number of signals code value is less than saturation threshold, then light leak number of signals code value is sent into subtracter, by subtracter with light leak number of signals code value from the original image pixels gray value of light leak signal same column deduct, obtain removing the view data behind the light leak signal; If light leak number of signals code value is greater than saturation threshold, then be weighted view data after on average obtaining proofreading and correct by regional pixel count code value about weighted average arithmetic element utilization generation light leakage phenomena zone.
6. the real-time cancellation element of transition surface array CCD camera light leakage phenomena between spatial row according to claim 5 is characterized in that described weighted average arithmetic element utilizes the view data after formula (4) obtains removing the light leak signal;
Wherein n is the columns of the zone of the light leakage phenomena pixel of occupying, then I
y(i, j), I
y(i, j+1) ..., I
y(i, j+n-1) the number value that obtains after calibrated for each row pixel of zone of saturation, I
x(i, j-n), I
x(i, j-n-1) ..., I
x(i j-1) is got the digital value of n row pixel original image, I by the left side, zone of saturation
x(i, j+n), I
x(i, j+n+1) ..., I
x(i, j+2n-1) by the right side, zone of saturation got the digital value of n row pixel original image,
For getting the weights of each row pixel on the left of the zone of saturation and right side;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110272780 CN102316283B (en) | 2011-09-15 | 2011-09-15 | Real-time light leak phenomenon removing device for space and line space transfer CCD (charge coupled device) camera |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110272780 CN102316283B (en) | 2011-09-15 | 2011-09-15 | Real-time light leak phenomenon removing device for space and line space transfer CCD (charge coupled device) camera |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102316283A CN102316283A (en) | 2012-01-11 |
CN102316283B true CN102316283B (en) | 2013-07-03 |
Family
ID=45429061
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201110272780 Expired - Fee Related CN102316283B (en) | 2011-09-15 | 2011-09-15 | Real-time light leak phenomenon removing device for space and line space transfer CCD (charge coupled device) camera |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102316283B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102739924B (en) * | 2012-05-31 | 2014-04-16 | 浙江大华技术股份有限公司 | Image processing method and system |
CN103344222B (en) * | 2013-06-24 | 2015-02-18 | 中国科学院长春光学精密机械与物理研究所 | Method for automatically adjusting space camera gain in real time |
CN104394411B (en) * | 2014-11-28 | 2018-01-26 | 上海集成电路研发中心有限公司 | Medium filtering device and method |
CN108259795B (en) * | 2018-01-17 | 2020-08-11 | 北京大恒图像视觉有限公司 | Image buffering method for spaced transfer type CCD |
CN109084893B (en) * | 2018-08-30 | 2020-11-10 | 江苏赛诺格兰医疗科技有限公司 | Light-shading detection device and method for photosensitive detector |
CN111062857B (en) * | 2019-11-25 | 2024-03-19 | 上海芯歌智能科技有限公司 | System and method for eliminating reflected light of 3D contour camera |
CN112672141B (en) * | 2020-12-16 | 2022-01-07 | 长光卫星技术有限公司 | TDI CCD non-photosensitive area light leakage rate detection system and method |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4656115B2 (en) * | 2007-09-27 | 2011-03-23 | ソニー株式会社 | Imaging apparatus, image signal processing circuit, image signal processing method, and computer program |
JP5226552B2 (en) * | 2009-02-03 | 2013-07-03 | オリンパスイメージング株式会社 | Imaging device |
CN101909144B (en) * | 2009-06-08 | 2012-06-27 | 华晶科技股份有限公司 | Image data correction method |
KR101375830B1 (en) * | 2009-11-16 | 2014-03-17 | 캐논 가부시끼가이샤 | Image capture apparatus and control method of same |
-
2011
- 2011-09-15 CN CN 201110272780 patent/CN102316283B/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN102316283A (en) | 2012-01-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102316283B (en) | Real-time light leak phenomenon removing device for space and line space transfer CCD (charge coupled device) camera | |
KR100315897B1 (en) | Adaptive non-uniformity compensation algorithm | |
CN109068025B (en) | Lens shadow correction method and system and electronic equipment | |
EP3509034A1 (en) | Image filtering based on image gradients | |
CN102917183A (en) | Imaging device, image processing method and program | |
CN104252700A (en) | Histogram equalization method for infrared image | |
CN105160657B (en) | Infrared imaging Nonuniformity Correction system based on FPGA | |
CN108304821A (en) | Image-recognizing method and device, image acquiring method and equipment, computer equipment and non-volatile computer readable storage medium storing program for executing | |
CN109255758A (en) | Image enchancing method based on full 1*1 convolutional neural networks | |
CN101571947A (en) | Image processing device and image processing method | |
DE102012221667A1 (en) | Apparatus and method for processing remote sensing data | |
CN114998141B (en) | Space environment high dynamic range imaging method based on multi-branch network | |
CN111738969A (en) | Image fusion method and device and computer readable storage medium | |
CN116596792B (en) | Inland river foggy scene recovery method, system and equipment for intelligent ship | |
CN115115549A (en) | Image enhancement model, method, equipment and storage medium of multi-branch fusion attention mechanism | |
CN110969584B (en) | Low-illumination image enhancement method | |
CN114693528A (en) | Unmanned aerial vehicle low-altitude remote sensing image splicing quality evaluation and redundancy reduction method and system | |
CN112712542B (en) | Foundation cloud picture motion prediction method combining block matching and optical flow method | |
CN113935917A (en) | Optical remote sensing image thin cloud removing method based on cloud picture operation and multi-scale generation countermeasure network | |
CN104065853B (en) | Infrared camera crosstalk eliminating method | |
CN106791803B (en) | A kind of disturbance measurement imaging system | |
Putra et al. | Visibility restoration of lake crater hazy image based on dark channel prior | |
US9210386B2 (en) | Filter setup learning for binary sensor | |
EP3002731A2 (en) | Determing color information using a binary sensor | |
CN110942425A (en) | Reconstruction method and reconstruction system of super-resolution image and electronic equipment |
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: 20130703 Termination date: 20140915 |
|
EXPY | Termination of patent right or utility model |