CN112750077A - Parallelized synthetic aperture radar image sub-band splicing effect processing method - Google Patents
Parallelized synthetic aperture radar image sub-band splicing effect processing method Download PDFInfo
- Publication number
- CN112750077A CN112750077A CN202011463637.6A CN202011463637A CN112750077A CN 112750077 A CN112750077 A CN 112750077A CN 202011463637 A CN202011463637 A CN 202011463637A CN 112750077 A CN112750077 A CN 112750077A
- Authority
- CN
- China
- Prior art keywords
- image
- sub
- band
- band splicing
- column
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/80—Geometric correction
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10044—Radar image
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a parallelized processing method for sub-band splicing effect of synthetic aperture radar images, which comprises the following steps: s1, reading in the image, and preprocessing the image to obtain an image to be processed; s2, processing the large-size image in a blocking mode; s3, parallelly processing the sub image blocks after the partitioning step; and S4, estimating the sub-band splicing error by parallelization calculation. The method can improve the processing effect of the algorithm when processing the sub-band splicing effect, reduce the error of image processing, and meanwhile, the parallelization calculation method can effectively improve the processing efficiency of the sub-band splicing effect and can be widely applied to the field of SAR image processing.
Description
Technical Field
The invention relates to the field of signal processing, in particular to a parallelized synthetic aperture radar image sub-band splicing effect processing method.
Background
The synthetic aperture radar is used as an active microwave sensor, has the characteristic of realizing all-weather earth observation all day long without limitation of illumination, weather conditions and the like, has wide application prospect in civil fields such as agriculture, forestry, oceans or geological and natural disasters, and has unique advantages in the military field.
Nowadays, the requirement of space-to-ground observation is continuously increasing, and in order to realize wide-area large-area observation, a ScanSAR (scanning synthetic aperture radar) and TOPS (top surface) working modes are developed by a satellite-borne SAR system. The working mode obtains larger imaging bandwidth by periodically adjusting the posture of the antenna. However, such an operation mode brings about the problems of non-uniform system gain and sub-band splicing errors, i.e., scallop effect and sub-band splicing effect. Although the TOPS mode has been effective in solving the scallop effect problem, transitions between sub-bands still exist. At present, a Kalman filtering-based method achieves a satisfactory effect in the aspect of correcting the sub-band splicing effect, but the method still has some problems. Firstly, if the orientation of the SAR image corresponds to the column of the image, the method takes the sub-band splicing error on each column of the image as a constant, and then uses a Kalman filter to estimate the constant error. On the other hand, the algorithm needs to estimate the subband error of each column of the image, the calculation amount is large, and the timeliness requirement is difficult to meet in practical application.
Disclosure of Invention
The invention aims to overcome the defects of the background technology, and provides a parallelized processing method for the sub-band splicing effect of a synthetic aperture radar image, so that the block processing idea is provided for solving the problem that the existing mathematical model is poor in fitting on a large image; aiming at the problem of computing timeliness, a thought for parallelizing the algorithm is provided; the invention can improve the processing effect of the algorithm when processing the sub-band splicing effect, reduce the error when processing the image, and simultaneously, the parallelization calculation method can effectively improve the processing efficiency of the sub-band splicing effect.
The invention provides a parallelized processing method for sub-band splicing effect of synthetic aperture radar images, which comprises the following steps: s1, reading in the image, and preprocessing the image to obtain an image to be processed; s2, processing the large-size image in a blocking mode; s3, parallelly processing the sub image blocks after the partitioning step; and S4, estimating the sub-band splicing error by parallelization calculation.
In the above technical solution, the specific steps of step S1 are as follows: s11, judging the stripe direction in the image, specifically judging whether the sub-band splicing error stripes, namely the azimuth direction of the SAR image, are parallel to the image and are distributed in a longitudinally arranged column, if not, transposing the image; and S12, determining the sub-band interval D.
In the above technical solution, the specific steps of step S2 are as follows: s21, calculating image size: length, width, M, N, i.e. the image has M rows and N columns; s22, judging whether the block is needed to be divided, if M <2000, directly entering the step S4; if M is greater than 2000, the image needs to be subjected to blocking processing; s23, dividing the columns of the image, wherein the image blocking method comprises the following steps: actual number of divided blocks s: s is Floor (M/1000), where Floor () is a Floor function, and if MOD (M/1000) is 0, the division result is s 1000 image blocks of size × N, and if MOD (M/1000) is x, the division result is s-1 image blocks of size 1000 × N and one image block of size (1000+ x) N.
In the above technical solution, the specific steps of step S3 are as follows: s31, determining the corresponding index position of the image block in the original image, and calculating the original Mean value [ Mean _1, Mean _2 … … Mean _ S ] of each image block]So as to facilitate image splicing after processing; s32, respectively performing sub-band splicing effect suppression on the image blocks by using a sub-band splicing effect processing algorithm based on Kalman filtering on the S image blocks, and equally dividing computing tasks to available computing nodes for distributed parallel computing; s33, mean value correction is carried out on the S image blocks to ensure that brightness jump does not occur at the spliced positions of the spliced image blocks, and the mean value correction method is as follows:
wherein Ioutput_x(m, n) and I'output_x(m, n) are respectively the pixel values of the (m, n) th image block after sub-band splicing effect suppression and the pixel values of the (m, n) th image block after Mean value adjustment, Mean _ x is the original Mean value of the x-th image block, and Output _ x _ Mean is the Mean value of the x-th image block after sub-band splicing effect suppression; and S34, re-splicing the S image blocks according to the index positions to obtain a final processing result.
In the above technical solution, the specific steps of step S32 are as follows: s321, calculating the mean and variance var of the neighborhood region of the current column, and splicing according to sub-bandsAn additive model of errors takes mean and var as parameters, and a Kalman filter is used for estimating a sub-band splicing error o corresponding to each column of pixels of an image blockS(ii) a S322, sub-band splicing error correction is carried out, and a sub-band splicing suppression result is obtained by directly subtracting the sub-band splicing error corresponding to a column in the original image; and S323, sequentially performing the steps S321 and S322 one by one until the whole image is processed, and obtaining a sub-band splicing inhibition result.
In the above technical solution, the specific steps of step S4 are as follows: s41, if the image does not need to be blocked, directly performing sub-band splicing effect processing based on a Kalman filter; s42, if the image is blocked, calculating parameters required by the Kalman filter; s43, distributing the current image matrix data calculated in the step S42 to a calculation node, and performing sub-band splicing error estimation based on a Kalman filter to obtain a sub-band splicing error estimation result obtained by each column of the image; and S44, correcting the original image according to the sub-band splicing error estimation result to obtain a final processing result, namely finishing the sub-band splicing effect suppression of the image.
In the above technical solution, the specific process of step S42 is as follows: s421, when a Kalman filter is used for sub-band error estimation, the calculation of input parameters is completed before distributed parallel calculation, the input parameters of the Kalman filter comprise the mean value of the left and right neighborhoods of the current estimation column, the variance of the left and right neighborhoods of the current estimation column and the observed value of the current column, wherein the calculation method of the domain mean value and the domain variance comprises the following steps: if the current column is the ith column, calculating the image from
Is listed in
The mean and variance of the area corresponding to the column are the neighborhood mean and the neighborhood variance of the ith column; s422, storing the neighborhood mean value and the neighborhood variance parameter corresponding to each row of pixels into the M +1 th row and the M +2 th row of each row, wherein each row of data of the image matrix contains the original imagePixel, neighborhood mean, and neighborhood variance.
In the above technical solution, in step S12, the subband splicing error stripe direction is parallel to the column, and the subband interval D is 3000; in step S2, M is 18742, N is 12776, and M >2000, and it is necessary to perform a blocking process to obtain 18 sub-images.
In the above technical solution, the distributed computation of step S3 adopts a computation architecture based on Hadoop, and the implementation steps are as follows: uploading data to be processed to an HDFS (Hadoop distributed File System), and then outputting data to be sub image blocks to be processed in a form of < key, value >, wherein the key is a position index corresponding to the image block, and the value corresponds to the image block to be processed; the Mapper acquires < key, value > data to be processed, performs sub-band splicing effect suppression based on Kalman filtering in step 32, outputs a processing result in the form of < key, value >, adjusts the mean value of the image in the reduce process, and performs splicing according to the index position to obtain the processing result of the whole image.
In the above technical solution, the step S4 is calculated based on a Hadoop framework, and the data to be processed is output in a form of < key, value >, where key is a position index of a current column in an image, and value is a vector formed by pixel values of the current column and a neighborhood mean neighborhood variance corresponding to the current column; then, the Mapper acquires < key, value > data to be processed, a Kalman filter is used for estimating sub-band splicing errors in the pixels of the column, the obtained estimation result is transmitted in a < key, value > form, at the moment, the key is the position index of the current column in the image, and the value is the sub-band splicing errors of the current column; and then, executing Reduce process, and correcting the sub-band splicing error in the original image according to the index position, thereby obtaining a final processing result, namely finishing the processing of the image sub-band splicing effect.
The parallelized processing method for the sub-band splicing effect of the synthetic aperture radar image has the following beneficial effects:
(1) the improved parallelization synthetic aperture radar sub-band splicing effect processing technology provided by the invention can weaken the adverse effect of the original algorithm sub-band splicing effect modeling defect on the processing result, and can effectively inhibit the sub-band splicing effect in the image.
(2) The improved parallelization calculation improvement related to the improved parallelization synthetic aperture radar sub-band splicing effect processing technology can adapt to the processing of images with various sizes, and the processing result of the original algorithm cannot be influenced.
(3) The improved parallelization synthetic aperture radar sub-band splicing effect processing technology provided by the invention has the advantages that the parallelization improvement is carried out on the sub-band splicing effect processing algorithm based on the Kalman filtering, and the processing efficiency of the algorithm can be effectively improved.
Drawings
FIG. 1 is a schematic overall flow chart of a parallelized SAR image sub-band stitching effect processing method according to the present invention;
FIG. 2 is a schematic diagram of image blocking and image blocking results in step two of the parallelized SAR image sub-band stitching effect processing method of the present invention;
FIG. 3 is a schematic flow chart of step three of the parallelized SAR image sub-band stitching effect processing method of the present invention;
FIG. 4 is a schematic flow chart of step four of the parallelized SAR image sub-band stitching effect processing method of the present invention;
FIG. 5 is a prior art scanning mode synthetic aperture radar image with sub-band stitching effect in an embodiment;
FIG. 6 is the results of FIG. 5 without the chunking direct process;
FIG. 7 is the result of the chunking process of FIG. 5;
fig. 8 is a diagram showing the comparison between the result of the blocking process of fig. 5 and the original result without the blocking process.
Detailed Description
The invention is described in further detail below with reference to the following figures and examples, which should not be construed as limiting the invention.
The improved parallelization processing technology for the image sub-band splicing effect of the synthetic aperture radar is based on a sub-band splicing effect processing algorithm based on a Kalman filter, solves the problem of poor fitting performance of an additive mathematical model in an original algorithm in a large-size image by using the idea of image blocking processing, and improves the processing efficiency of the sub-band splicing effect by performing parallel processing on a plurality of image blocks; in addition, the invention carries out parallelization improvement on the step of sub-band splicing error estimation in the original algorithm, so that the processing efficiency of the image without block processing can be improved.
The invention discloses a parallelized processing method for sub-band splicing effect of synthetic aperture radar images, which has a flow chart shown in figure 1 and specifically comprises the following steps:
the method comprises the following steps: reading in an image, and preprocessing the image to obtain an image to be processed;
this step includes two parts:
(1) judging the stripe direction in the image, specifically judging whether the sub-band splicing error stripes, namely the azimuth direction of the SAR image, are distributed in parallel to the columns of the image, if not, transposing the image;
(2) a subband spacing D is determined.
The implementation condition of this step is that the input image is required to be an amplitude image or a power image.
In this embodiment, the subband direction is parallel to the column, and no transposition is required, and the subband interval D only needs one approximate range, where D is selected to be 3000 in this embodiment;
step two: blocking processing for large-size images;
checking the size of the image, wherein if the length of the image exceeds a threshold value, the original algorithm mathematical model cannot accurately fit sub-band splicing error distribution in the image, if the model is directly applied to suppress the sub-band splicing effect, artificial artifacts are generated in a processing result, but the mathematical model still has good fitting performance on local images, and therefore the image needs to be blocked; if the image size is small, the mathematical model of the original algorithm can better fit the sub-band splicing error distribution in the image, and the step of image blocking is not needed, and the step four is directly performed.
The method comprises the following three steps:
(1) calculating the image size: length, width, M, N, i.e. the image has M rows and N columns;
(2) judging whether the block division is needed, if M <2000, directly entering step S4; if M is greater than 2000, the image needs to be subjected to blocking processing;
(3) the reason why the columns of the image are divided is that:
in the prior art, the mathematical model of the original additive model algorithm for the sub-band splicing error of each column of pixels in the image is as follows:
IS(m)＝IO(m)+oS m∈1,M
wherein IO(m) and IS(M) the ideal and actual values for the mth pixel in the column of images, and o if the image is large, i.e. M is largeSCan not be regarded as a constant, so that the image needs to be segmented, so that the model has better fitting performance.
In the present invention, the image blocking method is as follows:
actual number of divided blocks s:
s＝Floor(M/1000)
the Floor () is a Floor function, and if MOD (M/1000) ═ 0, the division result is s 1000 × N image blocks, and if MOD (M/1000) ═ x, the division result is s-1 1000 × N image blocks and one (1000+ x) N image block.
In this embodiment, a scanning mode synthetic aperture radar image of 18742 × 12776 is processed, M is 18742, N is 12776, and M >2000, so that a blocking process is required to obtain 18 sub-images, and a blocking diagram of the image and an image blocking result are shown in fig. 2.
Step three: parallel processing of sub image blocks;
and step two, obtaining s image blocks after the blocking step, wherein the sub-band splicing effect suppression operation aiming at each image block is mutually independent, so that parallel processing can be carried out, and then the image blocks are spliced into a complete image to obtain a final processing result.
Therefore, in this step, distributed parallel computation is performed on the plurality of sub-image blocks generated in step two, and finally, the sub-image blocks are re-spliced to obtain a final sub-band splicing effect suppression result.
The method can be divided into the following four steps:
(1) determining the corresponding index position of the image block in the original image, and calculating the original Mean value [ Mean _1, Mean _2 … … Mean _ s ] of each image block so as to facilitate image splicing after processing;
(2) respectively carrying out sub-band splicing effect suppression on the image blocks by using a sub-band splicing effect processing algorithm based on Kalman filtering on the s image blocks, and equally dividing computing tasks to available computing nodes for distributed parallel computing; the sub-band splicing effect processing steps based on Kalman filtering are as follows:
a) calculating mean and variance var of a neighborhood region of a current column, and estimating a sub-band splicing error o corresponding to each column of pixels of the image block by using a Kalman filter by taking mean and var as parameters according to a mathematical model of the sub-band splicing errorS。
b) And (4) correcting the sub-band splicing error, and directly subtracting the sub-band splicing error corresponding to a column in the original image from the column to obtain a sub-band splicing inhibition result.
c) And (3) sequentially carrying out the steps a) and b) one by one until the whole image is processed, and obtaining a sub-band splicing inhibition result.
(3) Performing mean value correction on the s image blocks to ensure that the splicing positions of the spliced image blocks do not have brightness jump; the mean value correction method comprises the following steps:
wherein Ioutput_x(m, n) and I'output_x(m, n) are respectively the pixel values of the (m, n) position in the x image block after sub-band splicing effect suppression and the x image block after Mean value adjustment, and Mean _ x is the x image blockThe Output _ x _ Mean is the Mean value of the xth image block after sub-band splicing effect suppression;
(4) and re-splicing the s image blocks according to the index positions to obtain a final processing result.
In this embodiment, a computing architecture based on Hadoop is adopted for distributed computing, a computing process is shown in fig. 3, data to be processed is uploaded to an HDFS during implementation, and then output data is a sub-image block to be processed in a form of < key, value >, where key is a position index corresponding to the image block and value corresponds to the image block to be processed; and (3) the Mapper acquires < key, value > data to be processed, sub-band splicing effect suppression based on Kalman filtering in the step (2) is carried out, then processing results are output in a < key, value > form, finally the mean value of the image is adjusted in the reduce process, and then splicing is carried out according to the index position, so that the processing result of the whole image is finally obtained.
Step four: parallelization calculation of sub-band splicing error estimation;
this step includes four parts:
(1) if the image does not need to be blocked, the sub-band splicing effect processing based on the Kalman filter is directly carried out, and the sub-band splicing error estimation steps of each column of pixels by the Kalman filter are considered to have better independence, so that the steps of the algorithm can be modified, and then the sub-band splicing error estimation can be subjected to parallelization distributed computation, so that the processing efficiency of the sub-band splicing effect is improved.
(2) Calculating parameters required by a Kalman filter;
in the process of using the Kalman filter to estimate the sub-band error, the input of the Kalman filter mainly comprises the mean value and the variance of the left and right neighborhoods of the current estimation column and the observed value of the current column. The observed value of the current column is the difference between the pixel value of the column of the image and the mean value of the neighborhood, so that the input parameters to be estimated by using the Kalman filter only comprise the mean value and the variance of the left neighborhood and the right neighborhood. In order to make the operation of each column of pixels completely independent, the calculation of input parameters is completed before the distributed parallel calculation is performed, and the neighborhood mean and the neighborhood variance parameters corresponding to each column of pixels are stored in the M +1 th row and the M +2 th row of each column, and at this time, each column of data of the image matrix comprises the original pixels of the image, the neighborhood mean and the neighborhood variance.
The method for calculating the domain mean value and the domain variance comprises the following steps: if the current column is the ith column, calculating the image from
Is listed in
And the mean and the variance of the area corresponding to the column are the neighborhood mean and the neighborhood variance of the ith column.
(3) Distributing current image matrix data to a computing node, and performing sub-band splicing error estimation based on a Kalman filter to obtain a sub-band splicing error estimation result obtained by each column of the image;
(4) and correcting the original image according to the sub-band splicing error estimation result to obtain a final processing result.
In this embodiment, the image is subjected to block processing, and parallel computation of each column is not involved. The step in the present invention can also be calculated based on a Hadoop architecture, and the flowchart is shown in fig. 4, and the data to be processed is also output in a form of < key, value >, where key is a position index of the current column in the image, and value is a vector composed of the pixel value of the current column and the neighborhood mean neighborhood variance corresponding to the current column. Then, the Mapper acquires < key, value > data to be processed, estimates the sub-band stitching error in the column of pixels by using a kalman filter, and transmits the obtained estimation result in a form of < key, value >, where key is the position index of the current column in the image and value is the sub-band stitching error of the current column. And then, executing a Reduce process, and correcting the sub-band splicing error in the original image according to the index position to obtain a final processing result.
And finishing the processing of the image sub-band splicing effect after the steps.
Examples
To illustrate the effectiveness of the present invention for algorithmic improvement, the following experiment was performed using a scan-mode synthetic aperture radar image of size 18742 x 12776 that includes the subband splicing effect (as shown in fig. 5). First, the original algorithm is used, and the block processing is not performed on it, and the obtained result is shown in fig. 6. Then, the result of processing using the method of image blocking processing for a large-size image mentioned in the present invention is shown in fig. 7. The detail magnification comparison results of the two images are shown in fig. 8.
As can be seen from fig. 5, there are significant brightness fluctuations at 3 at the location of the sub-band splice in the image. After the processing by the sub-band splicing effect suppression method based on the kalman filter, as shown in fig. 6, the brightness fluctuation in the image is corrected, however, due to the defect of the algorithm model, a plurality of obvious bright lines appear in the image, which is called artificial artifacts. In fig. 7, it is clearly seen that the sub-band stitching effect in the image is well corrected and no artifacts occur. Fig. 8 shows details of the processing result in an enlarged manner, and it can be seen that the adverse effect of the original algorithm subband splicing effect modeling defect on the processing result can be effectively weakened after the blocking processing is performed, and new errors cannot be brought to the image by splicing of the subsequent sub-images in the processing step.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Those not described in detail in this specification are within the skill of the art.
Claims (10)
1. A parallelized processing method for sub-band splicing effect of synthetic aperture radar images is characterized by comprising the following steps: the method comprises the following steps:
s1, reading in the image, and preprocessing the image to obtain an image to be processed;
s2, processing the large-size image in a blocking mode;
s3, parallelly processing the sub image blocks after the partitioning step;
and S4, estimating the sub-band splicing error by parallelization calculation.
2. The method for processing the sub-band stitching effect of the parallelized synthetic aperture radar image according to claim 1, wherein: the specific steps of step S1 are as follows:
s11, judging the stripe direction in the image, specifically judging whether the sub-band splicing error stripes, namely the azimuth direction of the SAR image, are parallel to the image and are distributed in a longitudinally arranged column, if not, transposing the image;
and S12, determining the sub-band interval D.
3. The method for processing the sub-band stitching effect of the parallelized synthetic aperture radar image according to claim 2, wherein: the specific steps of step S2 are as follows:
s21, calculating image size: length, width, M, N, i.e. the image has M rows and N columns;
s22, judging whether the block is needed to be divided, if M is less than 2000, directly entering the step S4; if M is larger than 2000, the image needs to be processed in a blocking mode;
s23, dividing the columns of the image, wherein the image blocking method comprises the following steps:
actual number of divided blocks s:
s＝Floor(M/1000)，
the Floor () is a Floor function, and if MOD (M/1000) ═ 0, the division result is s 1000 × N image blocks, and if MOD (M/1000) ═ x, the division result is s-1 1000 × N image blocks and one (1000+ x) N image block.
4. The method according to claim 3, wherein the method comprises the following steps: the specific steps of step S3 are as follows:
s31, determining the corresponding index position of the image block in the original image, and calculating the original Mean value [ Mean _1, Mean _2.. Mean _ S ] of each image block so as to facilitate image splicing after processing;
s32, respectively performing sub-band splicing effect suppression on the image blocks by using a sub-band splicing effect processing algorithm based on Kalman filtering on the S image blocks, and equally dividing computing tasks to available computing nodes for distributed parallel computing;
s33, mean value correction is carried out on the S image blocks to ensure that brightness jump does not occur at the spliced positions of the spliced image blocks, and the mean value correction method is as follows:
wherein Ioutput_x(m, n) and I'output_x(m, n) are respectively the pixel values of the (m, n) th image block after sub-band splicing effect suppression and the pixel values of the (m, n) th image block after Mean value adjustment, Mean _ x is the original Mean value of the x-th image block, and Output _ x _ Mean is the Mean value of the x-th image block after sub-band splicing effect suppression;
and S34, re-splicing the S image blocks according to the index positions to obtain a final processing result.
5. The method according to claim 4, wherein the method comprises the following steps: the specific steps of step S32 are as follows:
s321, calculating mean and variance var of the neighborhood region of the current column, and estimating a sub-band splicing error o corresponding to each column of pixels of the image block by using a Kalman filter according to an additive model of the sub-band splicing error and taking mean and var as parametersS；
S322, sub-band splicing error correction is carried out, and a sub-band splicing suppression result is obtained by directly subtracting the sub-band splicing error corresponding to a column in the original image;
and S323, sequentially performing the steps S321 and S322 one by one until the whole image is processed, and obtaining a sub-band splicing inhibition result.
6. The method according to claim 5, wherein the method comprises the following steps: the specific steps of step S4 are as follows:
s41, if the image does not need to be blocked, directly performing sub-band splicing effect processing based on a Kalman filter;
s42, if the image is blocked, calculating parameters required by the Kalman filter;
s43, distributing the current image matrix data calculated in the step S42 to a calculation node, and performing sub-band splicing error estimation based on a Kalman filter to obtain a sub-band splicing error estimation result obtained by each column of the image;
and S44, correcting the original image according to the sub-band splicing error estimation result to obtain a final processing result, namely finishing the sub-band splicing effect suppression of the image.
7. The method according to claim 6, wherein the method comprises the following steps: the specific process of step S42 is as follows:
s421, when a Kalman filter is used for sub-band error estimation, the calculation of input parameters is completed before distributed parallel calculation, the input parameters of the Kalman filter comprise the mean value of the left and right neighborhoods of the current estimation column, the variance of the left and right neighborhoods of the current estimation column and the observed value of the current column, wherein the calculation method of the domain mean value and the domain variance comprises the following steps: if the current column is the ith column, calculating the image from
Is listed in
The mean and variance of the area corresponding to the column are the neighborhood mean and the neighborhood variance of the ith column;
s422, storing neighborhood mean values and neighborhood variance parameters corresponding to each row of pixels into the M +1 th row and the M +2 th row of each row, wherein each row of data of the image matrix comprises original pixels of the image, the neighborhood mean values and the neighborhood variances.
8. The method according to claim 7, wherein the method comprises the following steps: in step S12, the subband splicing error stripe direction is parallel to the column, and the subband interval D is 3000;
in step S2, M is 18742, N is 12776, and M >2000, and it is necessary to perform a blocking process to obtain 18 sub-images.
9. The method according to claim 8, wherein the method comprises the following steps: the distributed computing of the step S3 adopts a computing architecture based on Hadoop, and the implementation steps are as follows: uploading data to be processed to an HDFS (Hadoop distributed File System), and then outputting a sub image block to be processed with the form of < key and value, wherein the key is a position index corresponding to the image block, and the value corresponds to the image block to be processed; the Mapper acquires the < key, value > data to be processed, performs sub-band splicing effect suppression based on Kalman filtering in step 32, outputs a processing result in the form of < key, value > and finally adjusts the mean value of the image in the reduce process, and then splices the image according to the index position to finally obtain the processing result of the whole image.
10. The method according to claim 9, wherein the method comprises the following steps: the step S4 is based on Hadoop architecture to calculate, and outputs the data to be processed in the form of < key, value, where key is the position index of the current column in the image, and value is the vector composed of the pixel value of the current column and the neighborhood mean neighborhood variance corresponding to the current column; then, the Mapper acquires data less than key and value to be processed, a Kalman filter is used for estimating sub-band splicing errors in the pixels of the column, the obtained estimation result is transmitted in a form of less than key and value, the key is the position index of the current column in the image, and the value is the sub-band splicing error of the current column; and then, executing Reduce process, and correcting the sub-band splicing error in the original image according to the index position, thereby obtaining a final processing result, namely finishing the processing of the image sub-band splicing effect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011463637.6A CN112750077B (en) | 2020-12-14 | 2020-12-14 | Parallelized synthetic aperture radar image sub-band splicing effect processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011463637.6A CN112750077B (en) | 2020-12-14 | 2020-12-14 | Parallelized synthetic aperture radar image sub-band splicing effect processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112750077A true CN112750077A (en) | 2021-05-04 |
| CN112750077B CN112750077B (en) | 2023-01-20 |
Family
ID=75649170
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011463637.6A Active CN112750077B (en) | 2020-12-14 | 2020-12-14 | Parallelized synthetic aperture radar image sub-band splicing effect processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112750077B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160109570A1 (en) * | 2013-02-08 | 2016-04-21 | Thales Alenia | High-Resolution Stripmap SAR Imaging |
| CN107092933A (en) * | 2017-04-24 | 2017-08-25 | 中国科学院遥感与数字地球研究所 | A Sea Ice Classification Method for Synthetic Aperture Radar Scanning Pattern Images |
| CN109035152A (en) * | 2018-05-23 | 2018-12-18 | 电子科技大学 | A kind of diameter radar image non-local mean filtering method |
| CN110533665A (en) * | 2019-09-03 | 2019-12-03 | 北京航空航天大学 | A kind of SAR image processing method inhibiting scallop effect and subband splicing effect |
| CN110749882A (en) * | 2019-10-24 | 2020-02-04 | 中国科学院电子学研究所 | Image Domain Scallop Suppression Method and System Based on Frequency Domain Filtering |
| CN111510657A (en) * | 2019-12-18 | 2020-08-07 | 中国船舶重工集团公司第七0九研究所 | Multi-path radar and photoelectric video comprehensive display method and system based on FPGA |
| CN111768337A (en) * | 2020-06-01 | 2020-10-13 | 中国科学院空天信息创新研究院 | An image processing method, device and electronic device |
-
2020
- 2020-12-14 CN CN202011463637.6A patent/CN112750077B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160109570A1 (en) * | 2013-02-08 | 2016-04-21 | Thales Alenia | High-Resolution Stripmap SAR Imaging |
| CN107092933A (en) * | 2017-04-24 | 2017-08-25 | 中国科学院遥感与数字地球研究所 | A Sea Ice Classification Method for Synthetic Aperture Radar Scanning Pattern Images |
| CN109035152A (en) * | 2018-05-23 | 2018-12-18 | 电子科技大学 | A kind of diameter radar image non-local mean filtering method |
| CN110533665A (en) * | 2019-09-03 | 2019-12-03 | 北京航空航天大学 | A kind of SAR image processing method inhibiting scallop effect and subband splicing effect |
| CN110749882A (en) * | 2019-10-24 | 2020-02-04 | 中国科学院电子学研究所 | Image Domain Scallop Suppression Method and System Based on Frequency Domain Filtering |
| CN111510657A (en) * | 2019-12-18 | 2020-08-07 | 中国船舶重工集团公司第七0九研究所 | Multi-path radar and photoelectric video comprehensive display method and system based on FPGA |
| CN111768337A (en) * | 2020-06-01 | 2020-10-13 | 中国科学院空天信息创新研究院 | An image processing method, device and electronic device |
Non-Patent Citations (1)
| Title |
|---|
| 叶珏磊，等: "基于GNSS/INS与激光雷达的缓存池建图", 《激光应用技术》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112750077B (en) | 2023-01-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | HS2P: Hierarchical spectral and structure-preserving fusion network for multimodal remote sensing image cloud and shadow removal | |
| EP2556490B1 (en) | Generation of multi-resolution image pyramids | |
| Ding et al. | Uncertainty-based thin cloud removal network via conditional variational autoencoders | |
| CN106934771B (en) | Infrared image stripe noise removing method based on local correlation | |
| JP2021533427A (en) | Reconstruction method of geostationary sea color satellite data by empirical orthogonal function decomposition method | |
| EP2927864B1 (en) | Image processing device and image processing method | |
| CN110533665B (en) | A SAR image processing method to suppress scallop effect and subband stitching effect | |
| CN111768337B (en) | Image processing method and device and electronic equipment | |
| CN111738929A (en) | Image processing method and device, electronic device, and storage medium | |
| CN109685879B (en) | Method, device, equipment and storage medium for determining texture distribution of multi-view images | |
| US20030138161A1 (en) | Method and apparatus for enhancing an image using a wavelet-based retinex algorithm | |
| CN102129669A (en) | Least square area network color-homogenizing method of aerial remote sensing image | |
| Zhu et al. | Low-light image enhancement network with decomposition and adaptive information fusion | |
| Ma et al. | SCT-CR: A synergistic convolution-transformer modeling method using SAR-optical data fusion for cloud removal | |
| Suthar et al. | A survey on various image inpainting techniques to restore image | |
| CN120655691A (en) | Endoscope monocular image depth estimation method and system | |
| CN112750077B (en) | Parallelized synthetic aperture radar image sub-band splicing effect processing method | |
| CN110223243A (en) | The tensor restorative procedure of non local self similarity and low-rank canonical based on tensor | |
| CN118037596A (en) | Methods, equipment, media and products for repairing highlight areas of UAV vegetation images | |
| Zhang et al. | Group-based sparse representation for Fourier ptychography microscopy | |
| CN120219173B (en) | Remote sensing image super-resolution reconstruction method, system, equipment and storage medium | |
| CN114972518A (en) | Real-time target pose tracking method and device, computer equipment and storage medium | |
| CN120070173A (en) | Large-angle swing scanning image resolution unification method | |
| CN117372307B (en) | Multi-unmanned aerial vehicle collaborative detection distributed image enhancement method | |
| CN117057994A (en) | A super-resolution reconstruction method for remote sensing images based on prior knowledge |
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 |