CN112967206A - Self-adaptive infrared image and video vertical line removing method based on image blocking - Google Patents
Self-adaptive infrared image and video vertical line removing method based on image blocking Download PDFInfo
- Publication number
- CN112967206A CN112967206A CN202110321322.6A CN202110321322A CN112967206A CN 112967206 A CN112967206 A CN 112967206A CN 202110321322 A CN202110321322 A CN 202110321322A CN 112967206 A CN112967206 A CN 112967206A
- Authority
- CN
- China
- Prior art keywords
- image
- noise
- vertical stripe
- vertical line
- deltanoise
- 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
- 238000000034 method Methods 0.000 title claims abstract description 40
- 230000000903 blocking effect Effects 0.000 title claims abstract description 17
- 238000001914 filtration Methods 0.000 claims abstract description 72
- 238000011156 evaluation Methods 0.000 claims abstract description 14
- 239000000126 substance Substances 0.000 claims description 19
- 238000012545 processing Methods 0.000 claims description 13
- 238000007781 pre-processing Methods 0.000 claims description 12
- 230000003044 adaptive effect Effects 0.000 claims description 10
- 238000012937 correction Methods 0.000 claims description 8
- 238000003709 image segmentation Methods 0.000 claims description 8
- 206010040925 Skin striae Diseases 0.000 claims description 6
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 9
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000003672 processing method Methods 0.000 abstract description 2
- 230000008030 elimination Effects 0.000 abstract 1
- 238000003379 elimination reaction Methods 0.000 abstract 1
- 238000000638 solvent extraction Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 8
- 230000000007 visual effect Effects 0.000 description 6
- 238000000605 extraction Methods 0.000 description 3
- 238000009499 grossing Methods 0.000 description 3
- 238000003331 infrared imaging Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
Images
Classifications
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration by the use of more than one image, e.g. averaging, subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
Abstract
The invention relates to an infrared digital image processing method, in particular to a self-adaptive infrared image and video vertical stripe removing method based on image blocking. The invention aims to solve the technical problems that the traditional infrared vertical stripe removing method has large calculated amount, is not beneficial to hardware implementation or has vertical stripe residue and image edge side effect. The method comprises the steps of obtaining preprocessed image frame data, filtering the image data, partitioning images before and after filtering respectively, carrying out self-adaptive image edge evaluation by utilizing one-dimensional noise characteristic distribution of each image block in a column direction to obtain a vertical line noise superposition quantity, correcting a median value taken by taking a column as a unit to serve as a final vertical line noise superposition quantity of the column, and then carrying out vertical line elimination on an original image by taking the column as a unit by utilizing the final vertical line noise superposition quantity. The method can inhibit edge side effects brought by the traditional algorithm, is low in algorithm complexity, and has high instantaneity and engineering realizability.
Description
Technical Field
The invention relates to an infrared digital image processing method, in particular to a self-adaptive infrared image and video vertical stripe removing method based on image blocking.
Background
In an infrared imaging system, the same column of pixels in a reading circuit of an infrared detector usually share the same output circuit, and due to the fact that bias voltages of the column output circuit and the row output circuit are not completely consistent, non-uniform noise with vertical stripes as main characteristics, namely vertical stripe noise, can be generated in an image. Because the vertical line noise belongs to one of the expressions of infrared nonuniformity, the traditional calibration or scene-based correction method can generate a certain inhibition effect on the vertical line noise, but the residual vertical line noise still has a larger visual effect on the image.
Besides the traditional infrared correction algorithm, the existing methods for removing the infrared vertical line noise mainly have two types:
the first type is a transformation-based method of striae erecta removal. The method needs to transform the original image data into a Fourier space domain or a wavelet domain, then carries out calibration of the vertical line noise, and finally carries out inverse domain transformation to achieve the purpose of noise removal.
The second type is a method for removing striae erecta based on feature statistics. The method filters the global image, uses the high-frequency information difference as vertical stripe non-uniformity characteristic data, and then uses the original image and the vertical stripe characteristic data to make difference so as to achieve the purpose of removing the vertical stripes. Because the extraction result of the vertical stripe features directly influences the effect of vertical stripe removal, if the vertical stripe feature extraction is incomplete, the vertical stripe residue is caused; if the extraction of the vertical stripe feature is excessive, the side effect of the edge of the normal image can be caused, and an 'edge artifact' is formed.
Disclosure of Invention
The invention aims to solve the technical problems that the traditional infrared vertical stripe removing method has large calculated amount, is not beneficial to hardware implementation or has vertical stripe residue and image edge side effect, and provides a self-adaptive infrared image and video vertical stripe removing method based on image blocking.
In order to solve the technical problems, the technical solution provided by the invention is as follows:
the invention provides a self-adaptive infrared image vertical stripe removing method based on image blocking, which is characterized by comprising the following steps:
1) preprocessing the collected infrared image to primarily eliminate non-uniform noise and blind element points, and taking the preprocessed image as an input image p;
2) carrying out low-pass guide filtering processing on the input image p to smooth noise to obtain a filtering result image q;
3) carrying out the same transverse blocking processing on the input image p and the filtering result image q, wherein the number of blocks is n, and n is more than or equal to 2;
4) respectively calculating the average gray value of each column in each horizontal block of the input image p and the filtering result image q;
5) calculating the difference value of the average gray value of each row in each transverse block of the input image p and the average gray value of the corresponding row in the corresponding transverse block of the filtering result image q, and taking the difference value as the primary vertical stripe noise superposition amount DeltaNoise' of the row in the transverse block;
6) correcting the preliminary vertical stripe noise superposition amount DeltaNoise' by using a self-adaptive edge evaluation function to obtain a corrected vertical stripe noise superposition amount DeltaNoise, sequencing the corrected vertical stripe noise superposition amount DeltaNoise by taking the column as a unit according to the size, and selecting a middle value as the final vertical stripe noise superposition amount of the current frame image in the column;
7) and calculating the difference value of the final noise superposition amount of the input image p and the current frame image by taking the column as a unit to obtain the target infrared image for eliminating the vertical line noise.
Further, in step 2), the formula for the low-pass guided filtering is as follows:
wherein the content of the first and second substances,
E(ak,bk) Is the guide filtering error value;
akand bkAre all constant parameters;
ωka filtering window taking r as a radius in the image;
p is an input image;
i is a guide image;
epsilon is a regularization parameter;
i is omegakEach pixel point in the filtering window;
k is a filter window index;
σ2for guiding the image I in the filtering window omegakThe variance within.
Further, in step 6), the preliminary vertical line noise overlap amount DeltaNoise' is corrected by using the adaptive edge evaluation function, so as to obtain a corrected vertical line noise overlap amount DeltaNoise, which specifically includes:
6.1) continuously extracting the average gray values of M columns for each transverse block of the input image p by using the columns as a unit, wherein M is more than or equal to 3;
6.2) to
Performing convolution operation on the extracted M vertical streak noise superposition quantity data to obtain an image horizontal gradient Grad for a convolution kernel with the step length of 1, wherein the convolution formula is as follows:
wherein the content of the first and second substances,
j is 0 … col as a column index;
col is the number of image columns;
m-0 … 1 corresponds to the element index of the convolution kernel;
6.3) obtaining the vertical stripe noise variation trend to NoiseTrend by using the image horizontal gradient Grad:
NoiseTrend(j+1)＝Grad(j)*Grad(j+1)
6.4) based on the NoiseTrend value, preliminarily correcting Delta Noise':
6.5) setting 2 empirical thresholds as an empirical upper threshold and an empirical lower threshold for limiting the vertical line noise respectively to determine the intensity for removing the vertical line noise, and further correcting by using the empirical upper threshold and the empirical lower threshold to obtain a corrected vertical line noise superposition amount DeltaNoise:
wherein the content of the first and second substances,
DeltaThr1 is an empirical upper threshold;
DeltaThr2 is an empirical lower threshold.
Further, in step 1), the preprocessing mode is non-uniform correction and blind pixel compensation.
Meanwhile, the invention also provides a self-adaptive infrared video vertical stripe removing method based on image blocking, which is characterized by comprising the following steps:
1) preprocessing an N-1 frame image of the collected infrared video to preliminarily eliminate non-uniform noise and blind element points, and taking the preprocessed image as an input image p; the N-1 frame image is a previous frame image of the current frame image;
2) carrying out low-pass guide filtering processing on the input image p to smooth noise to obtain a filtering result image q;
3) carrying out the same transverse blocking processing on the input image p and the filtering result image q, wherein the number of blocks is n, and n is more than or equal to 2;
4) respectively calculating the average gray value of each column in each horizontal block of the input image p and the filtering result image q;
5) calculating the difference value of the average gray value of each row in each transverse block of the input image p and the average gray value of the corresponding row in the corresponding transverse block of the filtering result image q, and taking the difference value as the primary vertical stripe noise superposition amount DeltaNoise' of the row in the transverse block;
6) correcting the preliminary vertical stripe noise superposition amount DeltaNoise' by using a self-adaptive edge evaluation function to obtain a corrected vertical stripe noise superposition amount DeltaNoise, sorting the corrected vertical stripe noise superposition amount DeltaNoise by taking the column as a unit according to the size, and selecting an intermediate value as the final vertical stripe noise superposition amount of the N-1 frame image in the column;
7) the same steps from step 1) to step 6) are carried out on the Nth frame image of the infrared video to obtain the final vertical stripe noise superposition amount of each row of the Nth frame image; the Nth frame image is a current frame image;
8) weighting and combining the final vertical stripe noise overlapping quantity of the (N-1) th frame image and the final vertical stripe noise overlapping quantity of the (N) th frame image by taking the columns as units, calculating the difference value of the final vertical stripe noise overlapping quantity of the (N + 1) th frame image and the N (N) th frame image by taking the columns as units, and obtaining a target image of the (N + 1) th frame image for eliminating the vertical stripe noise in real time; and the (N + 1) th frame image is the next frame image of the current frame image.
Further, in step 2), the formula for the low-pass guided filtering is as follows:
wherein the content of the first and second substances,
E(ak,bk) Is the guide filtering error value;
akand bkAre all constant parameters;
ωka filtering window taking r as a radius in the image;
p is an input image;
i is a guide image;
epsilon is a regularization parameter;
i is omegakEach pixel point in the filtering window;
k is a filter window index;
σ2for guiding the image I in the filtering window omegakThe variance within.
Further, in step 6), the preliminary vertical line noise overlap amount DeltaNoise' is corrected by using a self-adaptive edge evaluation function, so as to obtain a corrected vertical line noise overlap amount DeltaNoise, which specifically includes:
6.1) continuously extracting the average gray values of M columns for each transverse block of the input image p by using the columns as a unit, wherein M is more than or equal to 3;
6.2) to
Performing convolution operation on the extracted M vertical streak noise superposition quantity data to obtain an image horizontal gradient Grad for a convolution kernel with the step length of 1, wherein the convolution formula is as follows:
wherein the content of the first and second substances,
j is 0 … col as a column index;
col is the number of image columns;
m-0 … 1 corresponds to the element index of the convolution kernel;
6.3) obtaining the vertical stripe noise variation trend to NoiseTrend by using the image horizontal gradient Grad:
NoiseTrend(j+1)＝Grad(j)*Grad(j+1)
6.4) based on the NoiseTrend value, preliminarily correcting Delta Noise':
6.5) setting 2 empirical thresholds as an empirical upper threshold and an empirical lower threshold for limiting the vertical line noise respectively to determine the intensity for removing the vertical line noise, and further correcting by using the empirical upper threshold and the empirical lower threshold to obtain a corrected vertical line noise superposition amount DeltaNoise:
wherein the content of the first and second substances,
DeltaThr1 is an empirical upper threshold;
DeltaThr2 is an empirical lower threshold.
Further, in step 1), the preprocessing mode is non-uniform correction and blind pixel compensation.
Compared with the prior art, the invention has the following beneficial effects:
1. the invention provides a self-adaptive infrared image and video vertical line removing method based on image blocking, which comprises the steps of obtaining preprocessed image frame data, filtering the image data, respectively blocking images before and after filtering, solving an average value by utilizing one-dimensional noise characteristic distribution of each image block in a column direction, carrying out self-adaptive image edge evaluation on the difference of the average values of the image blocks before and after filtering to obtain a corrected vertical line noise superposition quantity, taking a median value as a final vertical line noise superposition quantity of the column by using a column unit, and then carrying out vertical line removal on an original image by using the final vertical line noise superposition quantity in a column unit. The method can effectively eliminate visual influence caused by the vertical line noise on the premise of keeping original details, inhibit edge side effects caused by the traditional algorithm, has low algorithm complexity and high real-time performance and engineering realizability, can effectively eliminate the vertical line noise caused by a reading circuit commonly existing in infrared thermal imaging, and further improves the infrared imaging quality.
2. The image block-based self-adaptive vertical stripe removing method is suitable for infrared images and infrared videos. When the infrared image to be processed is non-video sequence data, the final vertical line noise superposition quantity is directly acted on the image, and the purpose of removing the vertical lines can be achieved. When a video sequence is processed, the weighted average value of the final vertical stripe noise overlapping amount of the current frame image and the final vertical stripe noise overlapping amount of the previous frame image is used for replacing the final vertical stripe noise overlapping amount of the current frame image, so that the data buffer storage amount is reduced, the real-time video playing frequency is improved, the purpose of vertical stripe removal can be achieved on the premise of not influencing the real-time playing frequency, the vertical stripe noise removal effect is not obviously reduced, the time domain smoothness is realized, the problem that the image gray value is changed due to the vertical stripe removal method of the video sequence, and visual light and shade alternation feeling is generated during video playing can be solved, the visual influence of vertical stripe noise removal on the video sequence is effectively inhibited, and the purpose of removing the real-time vertical stripe information of the infrared video sequence is achieved.
Drawings
FIG. 1 is a schematic main flow chart of an adaptive infrared video vertical streak removal method based on image segmentation according to the present invention;
fig. 2 is a detailed flowchart of the image-segmentation-based adaptive infrared video vertical streak removal method of the present invention.
Detailed Description
The invention is further described below with reference to the figures and examples.
For an infrared image, a self-adaptive infrared image vertical stripe removing method based on image blocks is adopted, and the method comprises the following steps:
1) preprocessing the acquired infrared image to primarily eliminate non-uniform noise and blind pixel points, wherein the preprocessing modes comprise but are not limited to non-uniform correction and blind pixel compensation, and the preprocessed image is used as an input image p of a striae removing method;
2) the method comprises the steps of conducting low-pass guide filtering processing on an input image p to smooth noise and weaken the embodiment of noise amount on numerical values to obtain a filtering result image q, wherein the guide filtering principle is to solve the optimal solution of a least square formula, namely under the premise that the input image and an expected output image are locally linearly related, the purpose of smoothing noise influence under the premise that the edge of the image is kept as much as possible is achieved by means of the filtering condition that the gradient directionality of the guide image and the expected output is consistent while the difference (namely the noise amount) between the minimum input and the expected output is made;
the formula for the low-pass guided filtering is as follows:
wherein the content of the first and second substances,
E(ak,bk) Is the guide filtering error value;
akand bkAre all constant parameters;
ωka filtering window taking r as a radius in the image;
p is an input image;
i is a guide image;
epsilon is a regularization parameter;
i is omegakEach pixel point in the filtering window;
k is a filter window index;
σ2for guiding the image I in the filtering window omegakVariance within;
3) carrying out the same transverse blocking processing on the input image p and the filtering result image q to obtain scene image blocks in different transverse areas so as to reduce the influence of a scene on the noise superposition quantity, wherein the number of the blocks is n, and n is more than or equal to 2; the n value can be adjusted along with the image resolution, and the larger the resolution is, the larger the n value is; each horizontal block of the input image p is Colmean [ i ], wherein i is 1 … n, and each horizontal block of the filtering result image q is a base map BaseImg [ i ], wherein i is 1 … n;
4) according to the vertical streak characteristic, the influence of the noise amount on the single element in each column is equivalent to the influence on the column mean, so that the average gray value of each column in each horizontal block of the input image p and the filtering result image q is calculated respectively, and the column mean of each horizontal block of the input image p is ColMean [ i ] [ j ], wherein i is 1 … n, and j is 1 … colum (total number of columns); the column mean value of each horizontal block base map BaseImg [ i ] of the filtering result image q is BColMean [ i ] [ j ], wherein i is 1 … n, and j is 1 … col (total column number);
5) calculating the difference value DeltaNoise '[ i ] [ j ], i is 1 … n, j is 1 … col (total columns) between the average gray value of each column in each transverse block of the input image p and the average gray value of the corresponding column in the corresponding transverse block of the filtering result image q, namely DeltaNoise' [ i ] [ j ]. Colmean [ i ] [ j ] -BColmean [ i ] [ j ], and taking the difference value as the preliminary vertical line noise superposition quantity of the column in the transverse block to extract high-frequency information containing the vertical line noise;
6) correcting the preliminary vertical line noise superposition amount DeltaNoise' i < j > by using a self-adaptive edge evaluation function, protecting image edge information, obtaining the corrected vertical line noise superposition amount DeltaNoise i < j >, sorting the corrected vertical line noise superposition amount DeltaNoise i < j > by taking the column as a unit (namely DeltaNoise i), and selecting an intermediate value as the final vertical line noise superposition amount of the current frame image in the column so as to obtain the noise superposition information with the minimum influence of a scene;
the method comprises the following steps of correcting the preliminary vertical line noise superposition amount DeltaNoise' by using a self-adaptive edge evaluation function to obtain a corrected vertical line noise superposition amount DeltaNoise, and specifically comprises the following steps:
6.1) continuously extracting the average gray values of M columns for each transverse block of the input image p by using the columns as a unit, wherein M is more than or equal to 3;
6.2) to
Performing convolution operation on the extracted M vertical streak noise superposition quantity data to obtain an image horizontal gradient Grad for a convolution kernel with the step length of 1, wherein the convolution formula is as follows:
wherein the content of the first and second substances,
j is 0 … col as a column index;
col is the number of image columns;
m-0 … 1 corresponds to the element index of the convolution kernel;
6.3) obtaining the vertical stripe noise variation trend to NoiseTrend by using the image horizontal gradient Grad:
NoiseTrend(j+1)＝Grad(j)*Grad(j+1)
6.4) based on the NoiseTrend value, preliminarily correcting Delta Noise':
6.5) setting 2 empirical thresholds as an empirical upper threshold and an empirical lower threshold for limiting the vertical line noise respectively to determine the intensity for removing the vertical line noise, and further correcting by using the empirical upper threshold and the empirical lower threshold to obtain a corrected vertical line noise superposition amount DeltaNoise:
wherein the content of the first and second substances,
DeltaThr1 is an empirical upper threshold;
delta Thr2 is an empirical lower threshold;
7) and calculating the difference value of the final noise superposition amount of the input image p and the current frame image by taking the column as a unit to obtain the target infrared image for eliminating the vertical line noise.
The infrared image to be processed is non-video sequence data, and the final vertical line noise superposition quantity is directly acted on the image, so that the purpose of removing the vertical lines can be achieved. When processing a video sequence, the method for removing the vertical stripes of the video sequence changes the gray value of the image, the video playing is easy to generate visual light and shade alternation, in order to effectively inhibit the visual influence of the vertical stripe removal on the video sequence, a time domain smoothing principle is required, namely, the weighted average value of the final vertical stripe noise superposition quantity of the current frame image and the last frame image is used for replacing the final vertical stripe noise superposition quantity of the current frame image, because if the current frame image is processed by using the final vertical stripe noise superposition quantity of the current frame image, the FPGA needs to buffer the current frame image data until the algorithm is completed, so that the buffer memory is increased, the video real-time playing frequency is reduced, and experiments prove that the final vertical stripe noise superposition quantity of the current frame image data is used for acting on the next frame image data, the purpose of removing the vertical stripes can be achieved on the premise of not influencing the real-time playing frequency, and does not significantly reduce the effect of removing the moire noise.
Therefore, for the infrared video, the method for removing the vertical stripes of the self-adaptive infrared video based on the image blocks is adopted, as shown in fig. 1 and 2, and comprises the following steps:
1) preprocessing an N-1 frame image of the acquired infrared video to preliminarily eliminate non-uniform noise and blind pixel points, wherein the preprocessing modes comprise but are not limited to non-uniform correction and blind pixel compensation, and the preprocessed image is used as an input image p; the N-1 frame image is a previous frame image of the current frame image;
2) the method comprises the steps of conducting low-pass guide filtering processing on an input image p to smooth noise and weaken the embodiment of noise amount on numerical values to obtain a filtering result image q, wherein the guide filtering principle is to solve the optimal solution of a least square formula, namely under the premise that the input image and an expected output image are locally linearly related, the purpose of smoothing noise influence under the premise that the edge of the image is kept as much as possible is achieved by means of the filtering condition that the gradient directionality of the guide image and the expected output is consistent while the difference (namely the noise amount) between the minimum input and the expected output is made;
the formula for the low-pass guided filtering is as follows:
wherein the content of the first and second substances,
E(ak,bk) Is the guide filtering error value;
akand bkAre all constant parameters;
ωka filtering window taking r as a radius in the image;
p is an input image;
i is a guide image;
epsilon is a regularization parameter;
i is omegakEach pixel point in the filtering window;
k is a filter window index;
σ2to guideImage I in the filter window omegakVariance within;
3) carrying out the same transverse blocking processing on the input image p and the filtering result image q to obtain scene image blocks in different transverse areas so as to reduce the influence of a scene on the noise superposition quantity, wherein the number of the blocks is n, and n is more than or equal to 2; the n value can be adjusted along with the image resolution, and the larger the resolution is, the larger the n value is; each horizontal block of the input image p is Colmean [ i ], wherein i is 1 … n, and each horizontal block of the filtering result image q is a base map BaseImg [ i ], wherein i is 1 … n;
4) according to the vertical streak characteristic, the influence of the noise amount on the single element in each column is equivalent to the influence on the column mean, so that the average gray value of each column in each horizontal block of the input image p and the filtering result image q is calculated respectively, and the column mean of each horizontal block of the input image p is ColMean [ i ] [ j ], wherein i is 1 … n, and j is 1 … colum (total number of columns); the column mean value of each horizontal block base map BaseImg [ i ] of the filtering result image q is BColMean [ i ] [ j ], wherein i is 1 … n, and j is 1 … col (total column number);
5) calculating the difference value DeltaNoise '[ i ] [ j ], i is 1 … n, j is 1 … colnum (total column number), namely DeltaNoise' [ i ] [ j ]. Colmean [ i ] [ j ] -BColmean [ i ] [ j ], of the average gray value of each column in each transverse block of the input image p and the average gray value of the corresponding column in the transverse block of the filtering result image q, and taking the difference value as the preliminary vertical line noise superposition quantity of the column in the transverse block to extract high-frequency information containing the vertical line noise;
6) correcting the preliminary vertical line noise superposition amount DeltaNoise' i < j > by using a self-adaptive edge evaluation function, protecting the edge information of the image, obtaining the corrected vertical line noise superposition amount DeltaNoise i < j >, sorting the corrected vertical line noise superposition amount DeltaNoise i < j > by taking the column as a unit (DeltaNoise i), and selecting a middle value as the final vertical line noise superposition amount of the N-1 frame image in the column to obtain the noise superposition information with the minimum influence of the scene;
in step 6), the preliminary vertical line noise superposition amount DeltaNoise' is corrected by using a self-adaptive edge evaluation function, so as to obtain a corrected vertical line noise superposition amount DeltaNoise, which specifically includes:
6.1) continuously extracting the average gray values of M columns for each transverse block of the input image p by using the columns as a unit, wherein M is more than or equal to 3;
6.2) to
Performing convolution operation on the extracted M vertical streak noise superposition quantity data to obtain an image horizontal gradient Grad for a convolution kernel with the step length of 1, wherein the convolution formula is as follows:
wherein the content of the first and second substances,
j is 0 … col as a column index;
col is the number of image columns;
m-0 … 1 corresponds to the element index of the convolution kernel;
6.3) obtaining the vertical stripe noise variation trend to NoiseTrend by using the image horizontal gradient Grad:
NoiseTrend(j+1)＝Grad(j)*Grad(j+1)
6.4) based on the NoiseTrend value, preliminarily correcting Delta Noise':
6.5) setting 2 empirical thresholds as an empirical upper threshold and an empirical lower threshold for limiting the vertical line noise respectively to determine the intensity for removing the vertical line noise, and further correcting by using the empirical upper threshold and the empirical lower threshold to obtain a corrected vertical line noise superposition amount DeltaNoise:
wherein the content of the first and second substances,
DeltaThr1 is an empirical upper threshold;
delta Thr2 is an empirical lower threshold;
7) the same steps from step 1) to step 6) are carried out on the Nth frame image of the infrared video to obtain the final vertical stripe noise superposition amount of each row of the Nth frame image; the Nth frame image is a current frame image;
8) weighting and combining the final vertical stripe noise superposition quantity of the N-1 th frame image and the final vertical stripe noise superposition quantity of the N-1 th frame image by using a column unit
The specific calculation process is as follows:
wherein the content of the first and second substances,
DeltaNoisePrethe final vertical stripe noise superposition quantity of the N-1 frame image is obtained;
DeltaNoiseThisthe final vertical line noise superposition quantity of the current Nth frame image;
DeltaNoiseNewnamely the final vertical line noise superposition quantity which is finally acted on the (N + 1) th frame image;
calculating the difference value of the final vertical streak noise superposition quantity of the (N + 1) th frame image and the N frame image by taking the column as a unit, and obtaining a target image of the (N + 1) th frame image for eliminating the vertical streak noise in real time; and the (N + 1) th frame image is the next frame image of the current frame image.
Finally, it should be noted that: the above embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same, and it is obvious for a person skilled in the art to modify the specific technical solutions described in the foregoing embodiments or to substitute part of the technical features, and these modifications or substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions protected by the present invention.
Claims (8)
1. A self-adaptive infrared image vertical stripe removing method based on image blocking is characterized by comprising the following steps:
1) preprocessing the collected infrared image to primarily eliminate non-uniform noise and blind element points, and taking the preprocessed image as an input image p;
2) carrying out low-pass guide filtering processing on the input image p to smooth noise to obtain a filtering result image q;
3) carrying out the same transverse blocking processing on the input image p and the filtering result image q, wherein the number of blocks is n, and n is more than or equal to 2;
4) respectively calculating the average gray value of each column in each horizontal block of the input image p and the filtering result image q;
5) calculating the difference value of the average gray value of each row in each transverse block of the input image p and the average gray value of the corresponding row in the corresponding transverse block of the filtering result image q, and taking the difference value as the primary vertical stripe noise superposition amount DeltaNoise' of the row in the transverse block;
6) correcting the preliminary vertical stripe noise superposition amount DeltaNoise' by using a self-adaptive edge evaluation function to obtain a corrected vertical stripe noise superposition amount DeltaNoise, sequencing the corrected vertical stripe noise superposition amount DeltaNoise by taking the column as a unit according to the size, and selecting a middle value as the final vertical stripe noise superposition amount of the current frame image in the column;
7) and calculating the difference value of the final noise superposition amount of the input image p and the current frame image by taking the column as a unit to obtain the target infrared image for eliminating the vertical line noise.
2. The image-segmentation-based adaptive infrared image striae removing method according to claim 1, wherein:
in step 2), the formula used by the low-pass guided filtering is as follows:
wherein the content of the first and second substances,
E(ak,bk) Is the guide filtering error value;
akand bkAre all constant parameters;
ωka filtering window taking r as a radius in the image;
p is an input image;
i is a guide image;
epsilon is a regularization parameter;
i is omegakEach pixel point in the filtering window;
k is a filter window index;
σ2for guiding the image I in the filtering window omegakThe variance within.
3. The image-segmentation-based adaptive infrared image striae removing method according to claim 1 or 2, wherein:
in step 6), the preliminary vertical line noise superposition amount DeltaNoise' is corrected by using the adaptive edge evaluation function, so as to obtain a corrected vertical line noise superposition amount DeltaNoise, which specifically includes:
6.1) continuously extracting the average gray values of M columns for each transverse block of the input image p by using the columns as a unit, wherein M is more than or equal to 3;
6.2) to
Performing convolution operation on the extracted M vertical streak noise superposition quantity data to obtain an image horizontal gradient Grad for a convolution kernel with the step length of 1, wherein the convolution formula is as follows:
wherein the content of the first and second substances,
j is 0 … col as a column index;
col is the number of image columns;
m-0 … 1 corresponds to the element index of the convolution kernel;
6.3) obtaining the vertical stripe noise variation trend to NoiseTrend by using the image horizontal gradient Grad:
NoiseTrend(j+1)＝Grad(j)*Grad(j+1)
6.4) based on the NoiseTrend value, preliminarily correcting Delta Noise':
6.5) setting 2 empirical thresholds as an empirical upper threshold and an empirical lower threshold for limiting the vertical line noise respectively to determine the intensity for removing the vertical line noise, and further correcting by using the empirical upper threshold and the empirical lower threshold to obtain a corrected vertical line noise superposition amount DeltaNoise:
wherein the content of the first and second substances,
DeltaThr1 is an empirical upper threshold;
DeltaThr2 is an empirical lower threshold.
4. The image-segmentation-based adaptive infrared image striae removing method according to claim 3, wherein:
in the step 1), the preprocessing mode is non-uniform correction and blind pixel compensation.
5. A self-adaptive infrared video vertical stripe removing method based on image blocking is characterized by comprising the following steps:
1) preprocessing an N-1 frame image of the collected infrared video to preliminarily eliminate non-uniform noise and blind element points, and taking the preprocessed image as an input image p; the N-1 frame image is a previous frame image of the current frame image;
2) carrying out low-pass guide filtering processing on the input image p to smooth noise to obtain a filtering result image q;
3) carrying out the same transverse blocking processing on the input image p and the filtering result image q, wherein the number of blocks is n, and n is more than or equal to 2;
4) respectively calculating the average gray value of each column in each horizontal block of the input image p and the filtering result image q;
5) calculating the difference value of the average gray value of each row in each transverse block of the input image p and the average gray value of the corresponding row in the corresponding transverse block of the filtering result image q, and taking the difference value as the primary vertical stripe noise superposition amount DeltaNoise' of the row in the transverse block;
6) correcting the preliminary vertical stripe noise superposition amount DeltaNoise' by using a self-adaptive edge evaluation function to obtain a corrected vertical stripe noise superposition amount DeltaNoise, sorting the corrected vertical stripe noise superposition amount DeltaNoise by taking the column as a unit according to the size, and selecting an intermediate value as the final vertical stripe noise superposition amount of the N-1 frame image in the column;
7) the same steps from step 1) to step 6) are carried out on the Nth frame image of the infrared video to obtain the final vertical stripe noise superposition amount of each row of the Nth frame image; the Nth frame image is a current frame image;
8) weighting and combining the final vertical stripe noise overlapping quantity of the (N-1) th frame image and the final vertical stripe noise overlapping quantity of the (N) th frame image by taking the columns as units, calculating the difference value of the final vertical stripe noise overlapping quantity of the (N + 1) th frame image and the N (N) th frame image by taking the columns as units, and obtaining a target image of the (N + 1) th frame image for eliminating the vertical stripe noise in real time; and the (N + 1) th frame image is the next frame image of the current frame image.
6. The image-segmentation-based adaptive infrared video vertical streak removal method of claim 5, wherein:
in step 2), the formula used by the low-pass guided filtering is as follows:
wherein the content of the first and second substances,
E(ak,bk) Is the guide filtering error value;
akand bkAre all constant parameters;
ωka filtering window taking r as a radius in the image;
p is an input image;
i is a guide image;
epsilon is a regularization parameter;
i is omegakEach pixel point in the filtering window;
k is a filter window index;
σ2for guiding the image I in the filtering window omegakThe variance within.
7. The image-segmentation-based adaptive infrared video vertical streak removal method according to claim 5 or 6, wherein:
in step 6), the preliminary vertical line noise superposition amount DeltaNoise' is corrected by using a self-adaptive edge evaluation function, so as to obtain a corrected vertical line noise superposition amount DeltaNoise, which specifically includes:
6.1) continuously extracting the average gray values of M columns for each transverse block of the input image p by using the columns as a unit, wherein M is more than or equal to 3;
6.2) to
Performing convolution operation on the extracted M vertical streak noise superposition quantity data to obtain an image horizontal gradient Grad for a convolution kernel with the step length of 1, wherein the convolution formula is as follows:
wherein the content of the first and second substances,
j is 0 … col as a column index;
col is the number of image columns;
m-0 … 1 corresponds to the element index of the convolution kernel;
6.3) obtaining the vertical stripe noise variation trend to NoiseTrend by using the image horizontal gradient Grad:
NoiseTrend(j+1)＝Grad(j)*Grad(j+1)
6.4) based on the NoiseTrend value, preliminarily correcting Delta Noise':
6.5) setting 2 empirical thresholds as an empirical upper threshold and an empirical lower threshold for limiting the vertical line noise respectively to determine the intensity for removing the vertical line noise, and further correcting by using the empirical upper threshold and the empirical lower threshold to obtain a corrected vertical line noise superposition amount DeltaNoise:
wherein the content of the first and second substances,
DeltaThr1 is an empirical upper threshold;
DeltaThr2 is an empirical lower threshold.
8. The image-segmentation-based adaptive infrared video vertical streak removal method of claim 7, wherein:
in the step 1), the preprocessing mode is non-uniform correction and blind pixel compensation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110321322.6A CN112967206B (en) | 2021-03-25 | 2021-03-25 | Self-adaptive infrared image and video vertical line removing method based on image blocking |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110321322.6A CN112967206B (en) | 2021-03-25 | 2021-03-25 | Self-adaptive infrared image and video vertical line removing method based on image blocking |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112967206A true CN112967206A (en) | 2021-06-15 |
| CN112967206B CN112967206B (en) | 2022-03-04 |
Family
ID=76278462
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110321322.6A Active CN112967206B (en) | 2021-03-25 | 2021-03-25 | Self-adaptive infrared image and video vertical line removing method based on image blocking |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112967206B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113538286A (en) * | 2021-07-29 | 2021-10-22 | 杭州微影软件有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN114286071A (en) * | 2021-12-09 | 2022-04-05 | 北京空间机电研究所 | Infrared image parity correction method based on block length optimization |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407920A (en) * | 2016-09-07 | 2017-02-15 | 深圳芯启航科技有限公司 | Stripe noise elimination method of fingerprint image |
| CN107767346A (en) * | 2017-09-08 | 2018-03-06 | 湖北久之洋红外系统股份有限公司 | A kind of infrared image fringes noise filtering method |
| CN109584204A (en) * | 2018-10-15 | 2019-04-05 | 上海途擎微电子有限公司 | A kind of image noise intensity estimation method, storage medium, processing and identification device |
| CN111161172A (en) * | 2019-12-18 | 2020-05-15 | 北京波谱华光科技有限公司 | Infrared image column direction stripe eliminating method, system and computer storage medium |
| CN111724315A (en) * | 2020-05-09 | 2020-09-29 | 中国人民解放军63686部队 | Infrared image noise removing method based on self-adaptive weighted median filtering |
| KR102187634B1 (en) * | 2019-06-27 | 2020-12-07 | 한화시스템 주식회사 | Apparatus for removing noise for infrared image |
-
2021
- 2021-03-25 CN CN202110321322.6A patent/CN112967206B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407920A (en) * | 2016-09-07 | 2017-02-15 | 深圳芯启航科技有限公司 | Stripe noise elimination method of fingerprint image |
| CN107767346A (en) * | 2017-09-08 | 2018-03-06 | 湖北久之洋红外系统股份有限公司 | A kind of infrared image fringes noise filtering method |
| CN109584204A (en) * | 2018-10-15 | 2019-04-05 | 上海途擎微电子有限公司 | A kind of image noise intensity estimation method, storage medium, processing and identification device |
| KR102187634B1 (en) * | 2019-06-27 | 2020-12-07 | 한화시스템 주식회사 | Apparatus for removing noise for infrared image |
| CN111161172A (en) * | 2019-12-18 | 2020-05-15 | 北京波谱华光科技有限公司 | Infrared image column direction stripe eliminating method, system and computer storage medium |
| CN111724315A (en) * | 2020-05-09 | 2020-09-29 | 中国人民解放军63686部队 | Infrared image noise removing method based on self-adaptive weighted median filtering |
Non-Patent Citations (1)
| Title |
|---|
| 张盛伟 等: "基于引导滤波的红外图像条纹噪声去除方法", 《计算机辅助设计与图形学学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113538286A (en) * | 2021-07-29 | 2021-10-22 | 杭州微影软件有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN114286071A (en) * | 2021-12-09 | 2022-04-05 | 北京空间机电研究所 | Infrared image parity correction method based on block length optimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112967206B (en) | 2022-03-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112967206B (en) | Self-adaptive infrared image and video vertical line removing method based on image blocking | |
| WO2011011445A1 (en) | System and method for random noise estimation in a sequence of images | |
| CN109584185B (en) | Image processing method | |
| US7092580B2 (en) | System and method using edge processing to remove blocking artifacts from decompressed images | |
| CN109325498B (en) | Vein extraction method for improving Canny operator based on window dynamic threshold | |
| CN106651792B (en) | Method and device for removing stripe noise of satellite image | |
| CN102819827A (en) | Self-adaption moment matching stripe noise removing method based on gray-level segmentation | |
| Jin et al. | Perceptual image quality assessment using block-based multi-metric fusion (BMMF) | |
| CN109447920B (en) | Band noise automatic removing method based on notch filter | |
| CN113192059A (en) | On-orbit ship detection method based on infrared remote sensing image | |
| CN110349106B (en) | Wavelet soft threshold image denoising method based on Renyi entropy | |
| RU2367015C1 (en) | Method of enhancing digital images | |
| CN111311508B (en) | Noise reduction method for pavement crack image with noise | |
| Hu et al. | A region-based video de-noising algorithm based on temporal and spatial correlations | |
| CN115546047A (en) | Video image noise reduction method, device and medium based on improved local filtering algorithm | |
| CN115797327A (en) | Defect detection method and device, terminal device and storage medium | |
| CN110751603A (en) | Method and system for enhancing image contrast and terminal equipment | |
| CN111652820B (en) | Method, device and image processing system for correcting non-uniformity of infrared image stripes | |
| Wang et al. | A bilateral filtering based ringing elimination approach for motion-blurred restoration image | |
| CN112991240B (en) | Image self-adaptive enhancement algorithm for real-time image enhancement | |
| CN113284056A (en) | Method for enhancing infrared image on water | |
| BV et al. | Image De-hazing techniques for Vision based applications-A survey | |
| Wang et al. | Adaptive Bright and Dark Channel Combined with Defogging Algorithm Based on Depth of Field | |
| CN114004757B (en) | Method, system, device and storage medium for removing interference in industrial image | |
| Li et al. | Background Suppression Method of Star Image Based on Improved CBDNet |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |