CN112465846A - Cloud-containing remote sensing image compression method based on filling strategy - Google Patents
Cloud-containing remote sensing image compression method based on filling strategy Download PDFInfo
- Publication number
- CN112465846A CN112465846A CN202011350894.9A CN202011350894A CN112465846A CN 112465846 A CN112465846 A CN 112465846A CN 202011350894 A CN202011350894 A CN 202011350894A CN 112465846 A CN112465846 A CN 112465846A
- Authority
- CN
- China
- Prior art keywords
- cloud
- image
- region
- mask
- remote sensing
- 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 62
- 238000007906 compression Methods 0.000 title claims abstract description 30
- 230000006835 compression Effects 0.000 title claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims description 27
- 230000008569 process Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 230000009466 transformation Effects 0.000 claims description 6
- 238000005260 corrosion Methods 0.000 claims description 3
- 230000007797 corrosion Effects 0.000 claims description 3
- 230000000873 masking effect Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000013139 quantization Methods 0.000 claims description 3
- 239000004576 sand Substances 0.000 claims description 3
- 230000008520 organization Effects 0.000 claims description 2
- 238000013467 fragmentation Methods 0.000 claims 1
- 238000006062 fragmentation reaction Methods 0.000 claims 1
- 238000001914 filtration Methods 0.000 abstract description 3
- 238000011084 recovery Methods 0.000 abstract description 3
- 238000012360 testing method Methods 0.000 description 12
- 238000004088 simulation Methods 0.000 description 10
- 230000002829 reductive effect Effects 0.000 description 5
- 230000003044 adaptive effect Effects 0.000 description 4
- 238000012669 compression test Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000036961 partial effect Effects 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000009499 grossing Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013144 data compression Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- MTHSVFCYNBDYFN-UHFFFAOYSA-N diethylene glycol Chemical compound OCCOCCO MTHSVFCYNBDYFN-UHFFFAOYSA-N 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
- G06T5/94—Dynamic range modification of images or parts thereof based on local image properties, e.g. for local contrast enhancement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- 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
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a cloud-containing remote sensing image compression method based on a filling strategy, and mainly solves the problems that in the prior art, more high-frequency information still exists at the edge of a cloud area after filling, and coding consumption is high. The scheme is as follows: carrying out boundary extension on the input cloud-containing image and the cloud mask image; calculating an approximate median of a region of interest in the cloud-containing image; acquiring a gray mask for marking a cloud area and a gray image for marking the edge of the cloud area; calculating the pixel mean value in the edge of each cloud area and filling the corresponding cloud area; carrying out edge mean filtering on the filled image; carrying out level shift on the filtered image by using the approximate median of the region of interest to obtain a preprocessed image; coding the preprocessed image to obtain a compressed code stream and sending the compressed code stream to a receiving end; and the receiving end decodes the received code stream to obtain a recovered image. The method can effectively reduce the coding rate of the cloud region, improve the recovery quality of the region of interest, and can be used for transmitting various cloud-containing images.
Description
Technical Field
The invention belongs to the technical field of image processing, and further relates to a method for efficiently compressing a remote sensing image containing an invalid cloud area, which can be used for transmitting and processing various remote sensing images containing clouds.
Background
The rapid development of high-resolution remote sensing technology makes the contradiction between satellite image transmission bandwidth and remote sensing image data volume continuously aggravated, and a rapid and efficient data compression algorithm is needed for transmitting and processing the acquired image. However, cloud regions which exist universally in the remote sensing images occupy a large amount of storage space and transmission bandwidth, and the bandwidth utilization rate of data ROI of the region of interest, namely terrain region data except the cloud regions in the images is reduced. Therefore, it is of great significance to research image compression algorithms containing invalid cloud region data. In many documents, several compression algorithms for images containing invalid regions have been proposed, which are mainly classified into two categories:
one class of methods is to modify the interior of the existing compression framework to achieve preferential coding of regions of interest or to skip over invalid data regions, which includes the MaxShift method, the Scaling-based method, the shape adaptive wavelet transform SA-DWT method, and the shape adaptive bit-plane coding SA-BPE method. The MaxShift method realizes the preferential coding of the interested area by improving the interested area coefficient or inhibiting the invalid area coefficient; the Scaling-based method realizes the preferential coding of the interested region by setting different priorities for the interested region; the shape adaptive wavelet transform SA-DWT method is improved by applying a mirror effect on the boundary of a data area; the shape adaptive bit-plane coding SA-BPE method avoids coding coefficients belonging to invalid data regions by skipping coefficients in a fractional bit-plane encoder at the Tier-1 coding level.
Another method is to pre-process the image as a front end of the existing compression framework to reduce the coding consumption of the invalid region as much as possible, which includes the Phagecyte algorithm, the ADR algorithm, and the LEC algorithm. The Phagecyte algorithm fills in invalid regions with the arithmetic mean of the pixels in the window of the designated region to adapt to local image variation, which is already applied in commercial GIS; the ADR algorithm fills the invalid region by adopting the average value of the interested region; the LEC algorithm fills the invalid region through the data average value in the local edge context, reduces the coding rate of the invalid region and simultaneously reduces the high-frequency information of the edge part.
Although the first method achieves the purpose of compressing invalid data areas by modifying the interior of the existing compression frame, the method is only suitable for processing images with interested areas in regular shapes, and for cloud-containing remote sensing images, the pixel blocks of the interested areas in the mapping process of each sub-band can be regarded as cloud pixel blocks due to the irregularity of the cloud areas, the coefficients of the interested areas are reduced while the coefficients of the cloud areas are restrained, and the compression performance is influenced.
Compared with the former method, the second method preprocesses the image before the existing compression frame, is easy to realize, has strong portability, and is insensitive to the shape of the region of interest. However, the current pretreatment methods still have defects, among them:
the invalid area filled by the Phagecyte algorithm is not a single gray value, and the coding consumption of the invalid area is still high; the filling value adopted by the ADR algorithm does not consider the edge information of the filling area, so that after wavelet transformation, more high-frequency information can be generated at the edge part, and the high-frequency information can generate higher coding consumption;
the LEC algorithm only considers the outer edge information, if the cloud detection algorithm does not detect the cloud of the edge, the pixel value of the filled cloud area is closer to the pixel value of the ground object area, and a larger pixel difference still exists between the filled cloud area and the cloud of the edge, so that higher coding consumption still exists at the edge.
Disclosure of Invention
The invention aims to provide a cloud-containing remote sensing image compression method based on a filling strategy aiming at the defects of the prior art, so that the cloud region coding consumption is reduced to the maximum extent and the compression performance of an interested region is improved by completely eliminating invalid cloud region data.
In order to achieve the purpose, the technical scheme of the invention comprises the following implementation steps:
1) expanding the boundary of the cloud-containing remote sensing image and the cloud mask image:
inputting a cloud-containing remote sensing image I and a corresponding cloud mask image M, wherein I is a gray image; m is a binary image, the region of interest is represented by 0, the cloud region is represented by 1, and the two images have the same size;
respectively complementing the width W and the height H of the cloud-containing remote sensing image I and the cloud mask image M into even numbers to obtain a cloud-containing remote sensing image I with an expanded boundarybAnd expanded cloud mask image Mb；
2) Calculating an approximate median of the region of interest:
respectively expanding the cloud-containing remote sensing image IbAnd expanded cloud mask image MbSpatially dividing the cloud mask image into a series of small square blocks with fixed sizes, and dividing the pixel points in the blocks of the cloud mask image into 0The block is regarded as the area of interest block;
cloud-containing remote sensing image I after calculation expansionbObtaining a mean value sequence A by taking the mean value of the pixel points in each interested region block, and taking the median of the sequence A as IbAn approximate median Z of the region of interest;
3) acquiring marked cloud area edges:
3a) for the expanded cloud mask image MbSequentially carrying out expansion and corrosion operations to obtain an expanded cloud mask MdAnd post-etch cloud mask mark Me；
3b) For the expanded cloud mask MdThe connected region marking is carried out on the cloud region, namely, each connected cloud region is represented by different numbers to obtain a gray image Nd(ii) a The expanded cloud mask MdAnd a post-etch cloud mask MePerforming point-by-point subtraction to obtain a binary image M with 1 representing cloud region edge and 0 representing other regionss；
3c) Masking the binary cloud with a mask MbAnd a gray scale image NdMultiplying point by point to obtain a gray mask N for marking different connected cloud areasb；
3d) The binary image MsAnd a gray scale image NdMultiplying point by point to obtain a gray image N marking the edges of different cloud regionss；
4) According to NsRespectively calculating and expanding cloud-containing remote sensing image I of the cloud region edge mark information inbStoring the result as a dictionary D according to the mean value of the pixels in the cloud region, and calculating the mean value of the pixels in the cloud region according to the gray mask NbThe cloud region mark information and the dictionary D in the Chinese cloud region are used for carrying out the cloud-containing remote sensing image I after the expansionbFilling each type of cloud area in the image to obtain a filled image If；
5) Traversing filled image IfIf a certain pixel point I is a pixel point in the interested areafThe four neighborhoods of (x, y) are internally provided with pixel points I belonging to cloud areasf(x ', y'), then IfThe value at the (x ', y') position is replaced by If(x ', y') and IfThe mean of the pixel values at (x, y);
6) filling the image IfSubtracting an offset from the values of all the pixels, wherein the value of the offset is equal to the median Z calculated in the step 2, and obtaining a preprocessed image Ip；
7) For the preprocessed image IpCarrying out compression coding to obtain a compressed code stream S and sending the compressed code stream S to a receiving end;
8) the receiving end decodes the received code stream S to obtain a recovered image Ir。
Compared with the prior art, the invention has the following beneficial effects:
first, the cloud area is filled according to the local inner and outer dual-edge context information, so that the problem that more high-frequency information still exists at the edge of the cloud area after filling in the prior art is solved, and the coding consumption is effectively reduced.
Secondly, the invention carries out mean filtering on the edges of the cloud region of the filled image, can further smooth the boundary and fully reduce the high-frequency wavelet coefficient at the edge.
Simulation results show that the cloud region coding rate can be reduced to the maximum extent, and the interesting region recovery quality is improved. Compared with a standard JPEG2000 algorithm, the peak signal-to-noise ratio PSNR index of the region of interest is obviously improved, and the higher the cloud content of the image is, the larger the index improvement is during the same-time compression.
Drawings
FIG. 1 is a general flow chart of an implementation of the present invention;
FIG. 2 is a sub-flowchart of the present invention for obtaining marked cloud region edges;
FIG. 3 is a schematic illustration of filling a cloud region in accordance with the present invention;
fig. 4 is a comparison graph of smoothing effect of the present invention and the prior art method on the cloud region boundary.
Detailed Description
The following further describes embodiments of the present invention with reference to the accompanying drawings:
referring to fig. 1, the specific implementation steps of the present invention are as follows:
step 1: and expanding the boundary of the cloud-containing remote sensing image and the cloud mask image.
1.1) inputting a cloud-containing remote sensing image I and a corresponding cloud mask image M, wherein I is a gray level image, and the bit depth is BN(ii) a M is a binary image, the region of interest is represented by 0, the cloud region is represented by 1, and the two images have the same size;
1.2) carrying out boundary extension on the cloud-containing remote sensing image:
if W is an odd number and H is an even number, adding a row of data to the rightmost side of the original cloud-containing remote sensing image, wherein the added data is the W-th row of data of the original cloud-containing remote sensing image;
if W is an even number and H is an odd number, adding a line of data at the lowest part of the original cloud-containing remote sensing image, wherein the added data is the H-th data of the original cloud-containing remote sensing image;
if W is an odd number and H is an odd number, firstly adding a column of data on the rightmost side of the original cloud-containing remote sensing image, wherein the added data is the W-th column of data of the original cloud-containing remote sensing image, recording the image filled with the column as a graph P, and then adding a row of data on the bottommost side of the graph P, wherein the added data is the H-th row of data of the graph P;
if W is an even number and H is an even number, the original cloud picture is not operated;
marking the cloud picture after the boundary expansion as Ib；
1.3) carrying out boundary extension on the cloud mask image:
the boundary extension process of the cloud mask image is completely consistent with that of the cloud remote sensing image, and the cloud mask image after the boundary extension is recorded as Mb；
1.4) cloud-containing remote sensing image I after boundary expansionbAnd expanded cloud mask image MbThe width is W '═ W + (W% 2) and the height is H' ═ H + (H% 2), where% indicates the modulus operation.
Step 2: an approximate median of the region of interest is calculated.
2.1) combining the expanded cloud mask MbFor extended cloud picture IbPartitioning:
respectively expanding the cloud-containing remoteImage sensing devicebAnd expanded cloud mask MbThe method comprises the following steps of spatially dividing the square blocks into a series of small square blocks with the same size, wherein the side length of each small square block can be selected from 8, 4 and 2;
if in fig. IbThe width and the height of the square can be completely divided by 8, and the side length of the small square block is set to be 8;
if in fig. IbThe width and the height of the square can not be divided by 8 but can be divided by 4, and the side length of the small square block is set to be 4;
if in fig. IbThe width and the height of the square can not be divided by 4 but can be divided by 2, and the side length of the small square block is set to be 2;
drawing MbThe blocks with the pixel values of all 0 in the blocks are regarded as region-of-interest blocks, and the total number of the region-of-interest blocks is recorded as m;
2.2) calculation of the graph IbObtaining the average value sequence A of the average values of the pixel points in each interested area block, wherein A is { a ═ a }1,a2,…,ai,…,amIn which a isiMeans representing the ith region of interest block, i ∈ [1, m]；
2.3) calculating the approximate median of sequence A as IbApproximate median of the region of interest:
opening up an array B ═ B0,b1,…,bj,…,b127Length 128, initialized to 0;
traversing the sequence A to connect the element aiIs divided by the value of
The result of (c) is denoted as w, and the element B in the array B iswAdding 1;
and sequentially accumulating the elements in the array B according to subscripts, and recording the result as:
judging the result of each accumulation, and if the Tth accumulated value is Sum(T)If m/2 is less, continue to accumulate and solve Sum(T+1)；
If Sum(T)Not less than m/2, thenFinish the accumulation and will
The value of (d) is taken as the approximate median of the region of interest and is denoted as Z.
And step 3: and acquiring marked cloud area edges.
Referring to fig. 2, the specific implementation of this step is as follows:
3.1) for the extended cloud mask image MbSequentially performing expansion and corrosion operations according to the 3 multiplied by 3 structural elements to obtain an expanded cloud mask MdAnd post-etch cloud mask mark Me；
3.2) cloud mask M after expansiondMarking the connected domain of the cloud region in the middle, and marking MdThe cloud region pixels conforming to 8-neighborhood connection are represented by the same reference number to obtain a gray image NdThe pixel value range of the gray level image is {0,1,2, …, l, …, n }, wherein 0 marks an interested area, l marks the first connected cloud area, and n represents the total number of the connected cloud areas;
3.3) masking the expanded cloud with a mask MdAnd the etched cloud mask MePerforming point-by-point subtraction to obtain a binary cloud edge image MsWherein 1 represents a cloud region edge, and 0 represents other regions;
3.4) expanded cloud mask image MbAnd a grayscale image NdMultiplying point by point to obtain a gray mask N for marking different connected cloud areasb，NbA set omega of pixel points with n non-zero gray values, wherein the gray value is kkForm the k-th cloud region, omegak＝{Nb(i,j)|Nb(i,j)＝k}，k∈[1,n]N is the number of the cloud area;
3.5) binary cloud edge image MsAnd a grayscale image NdMultiplying point by point to obtain a gray image N marking the edges of different cloud regionss，NsSet Γ of pixels with n non-zero gray values, where the gray value is kkConstituting class k cloud edges, Γk＝{Ns(i,j)|Ns(i,j)＝k}，k∈[1,n]And n is the number of the cloud area types.
And 4, step 4: and (5) filling the cloud area.
4.1) from the grayscale image NsRespectively calculating the cloud picture I after the boundary expansionbThe mean values of the pixels in the corresponding area are stored as a dictionary D ═ { D {1,D2,…,Dh,…,DnIn which D ishRepresenting the edge pixel mean value of the h-th cloud area, h belongs to [1, n ∈]；
4.2) according to graph NbThe region marking information and dictionary D in (1), the graph IbThe h-type cloud area in (1) is filled with DhAnd the filled image is marked as IfAs shown in fig. 3.
And 5: and filtering the edge mean value.
According to cloud mask MbGo through the filled image IfFor all the interested region pixel points, for each interested region pixel point If(x, y), judging whether cloud area pixel points exist in four neighborhoods of the image, namely the upper neighborhood, the lower neighborhood, the left neighborhood and the right neighborhood, if the four neighborhoods contain the cloud area pixel points If(x ', y'), then the cloud area pixel point If(x ', y') the pixel value is replaced by If(x ', y') and IfThe mean of the pixel values at (x, y).
Step 6: the level is shifted.
Before an image is sent to a compression core, direct current components in an image signal need to be eliminated as much as possible, so that after image transformation, the probability of positive and negative values of wavelet coefficients is basically the same, and the coding efficiency of a compression algorithm is improved.
Aiming at the characteristics of the cloud-containing remote sensing image, the embodiment takes the number of bits in the region of interest as the offset of level shift, namely the approximate median Z of the region of interest obtained by calculation in the step 2 as the offset, and takes the filled image IfSubtracting Z from the values of all the pixel points to complete the level shift, and obtaining a preprocessed image Ip。
And 7: and (5) compression coding.
Can adopt but is not limited to JAlgorithm of PEG2000 or JPEG-LS to the preprocessed image IpPerforming compression coding, wherein the example adopts a JPEG2000 algorithm to perform coding, and the coding comprises component transformation, fragment processing, wavelet transformation, quantization, entropy coding, code rate control and code stream organization processes in sequence; if the JPEG-LS algorithm is adopted for compression coding, the method sequentially comprises the processes of context modeling, prediction, residual calculation, parameter updating and Golomb-Rice coding. And coding to obtain a code stream S and sending the code stream S to a receiving end.
And 8: and (5) image decoding.
Decoding the received code stream S by adopting a decoding algorithm matched with the code of the sending end, in the embodiment, decoding by adopting JPEG2000, and obtaining a recovery image I by sequentially carrying out the processes of code stream decomposition, entropy decoding, inverse quantization, inverse wavelet transform and post-processingr。
The effects of the present invention can be further illustrated by the following simulations:
1. simulation conditions are as follows:
the remote sensing image shot by the China high score number 1 (GF-1) optical satellite is used for testing.
The image size is 1024 x 1024 with an accuracy of 10 bits.
The number of the test images was 150 in total, and the test images were divided into 3 groups of 50 images each based on the average cloud content.
The existing methods used for simulation include the following three methods:
firstly, a standard JPEG2000 algorithm is adopted;
second, ADR + JPEG2000 method for compressing by JPEG2000 algorithm after preprocessing test image by ADR algorithm
Thirdly, the LEC + JPEG2000 method is adopted to pre-process the test image by the LEC algorithm and then compress by the JPEG2000 algorithm
2. Simulation content:
table 1 test results for average cloud content 30% images
Simulation 2, performing compression test on the test image with average cloud content of 50% by using the invention and the above 3 existing methods, and counting the average value of peak signal-to-noise ratios (PSNR) of the interested areas of the methods under different compression multiples, wherein the result is shown in Table 2, and the unit is dB:
table 2 test results for average cloud 50% image
Simulation 3, performing compression test on the test image with the average cloud content of 70% by using the invention and the above 3 existing methods, and counting the average value of peak signal-to-noise ratios (PSNR) of the interested areas of the methods under different compression multiples, wherein the result is shown in Table 3, and the unit is dB:
table 3 test results for average cloud content 70% image
Simulation 4, performing compression test on a certain test image by using the invention and the ADR + JPEG2000 method and the LEC + JPEG2000 method, and comparing the smoothness of the cloud area boundaries by using the methods, wherein the result is shown in FIG. 4, wherein FIG. 4(a) is a cloud-containing remote sensing image, FIG. 4(b) is a cloud mask image, FIG. 4(c) is a partial enlarged view of the cloud area boundary of a certain position in the figure after the ADR + JPEG2000 method is adopted, FIG. 4(d) is a partial enlarged view of the cloud area boundary of a certain position in the figure after the LEC + JPEG2000 method is adopted, and FIG. 4(e) is a partial enlarged view of the cloud area boundary of a certain position in the figure after the ADR + JPEG2000 method is adopted.
3. Simulation conclusion
The data in table 1, table 2 and table 3 all show that the PSNR index of the region of interest of the present invention is greatly higher than that of other methods under the same compression multiple; and the higher the cloud content of the image, the better the compression performance of the invention.
As can be seen from fig. 4, the smoothing effect at the cloud region boundary is significantly better than that of other methods after the method is adopted.
The above description is only one specific example of the present invention and should not be construed as limiting the invention in any way. It will be apparent to persons skilled in the relevant art that various modifications and changes in form and detail can be made therein without departing from the principles and arrangements of the invention, but these modifications and changes are still within the scope of the invention as defined in the appended claims.
Claims (10)
1. A cloud-containing remote sensing image compression method based on a filling strategy is characterized by comprising the following steps:
1) expanding the boundary of the cloud-containing remote sensing image:
inputting a cloud-containing remote sensing image I and a corresponding cloud mask image M, wherein I is a gray image; m is a binary image, the region of interest is represented by 0, the cloud region is represented by 1, and the two images have the same size;
respectively complementing the width W and the height H of the cloud-containing remote sensing image I and the cloud mask image M into even numbers to obtain a cloud-containing remote sensing image I with an expanded boundarybAnd expanded cloud mask image Mb；
2) Calculating the approximate median of the region of interest:
respectively expanding the cloud-containing remote sensing image IbAnd expanded cloud mask image MbThe method comprises the steps of spatially dividing the cloud mask image into a series of small square blocks with the same size, and regarding blocks with pixel points of 0 in the blocks of the cloud mask image as region-of-interest blocks;
cloud-containing remote sensing image I after calculation expansionbObtaining a mean value sequence A of the mean values of the pixel points in each interested region block, and obtaining the sequenceThe median of A is taken as IbAn approximate median Z of the region of interest;
3) acquiring marked cloud area edges:
3a) for the expanded cloud mask image MbSequentially carrying out expansion and corrosion operations to obtain an expanded cloud mask MdAnd post-etch cloud mask mark Me；
3b) For the expanded cloud mask MdThe connected region marking is carried out on the cloud region, namely, each connected cloud region is represented by different numbers to obtain a gray image Nd(ii) a The expanded cloud mask MdAnd a post-etch cloud mask MePerforming point-by-point subtraction to obtain a binary image M with 1 representing cloud region edge and 0 representing other regionss；
3c) Masking the binary cloud with a mask MbAnd a gray scale image NdMultiplying point by point to obtain a gray mask N for marking different connected cloud areasb；
3d) The binary image MsAnd a gray scale image NdMultiplying point by point to obtain a gray image N marking the edges of different cloud regionss；
4) According to NsRespectively calculating and expanding cloud-containing remote sensing image I of the cloud region edge mark information inbStoring the result as a dictionary D according to the mean value of the pixels in the cloud region, and calculating the mean value of the pixels in the cloud region according to the gray mask NbThe cloud region mark information and the dictionary D in the Chinese cloud region are used for carrying out the cloud-containing remote sensing image I after the expansionbFilling each type of cloud area in the image to obtain a filled image If；
5) Traversing filled image IfIf a certain pixel point I is a pixel point in the interested areafThe four neighborhoods of (x, y) are internally provided with pixel points I belonging to cloud areasf(x ', y'), then IfThe value at the (x ', y') position is replaced by If(x ', y') and IfThe mean of the pixel values at (x, y);
6) filling the image IfSubtracting an offset from the values of all the pixels, wherein the value of the offset is equal to the median Z calculated in the step 2, and obtaining a preprocessed image Ip；
7) For the preprocessed image IpCarrying out compression coding to obtain a compressed code stream S and sending the compressed code stream S to a receiving end;
8) the receiving end decodes the received code stream S to obtain a recovered image Ir。
2. The method according to claim 1, wherein 1) the cloud-containing remote sensing image I after boundary expansionbAnd expanded cloud mask image MbThe width is W '═ W + (W% 2) and the height is H' ═ H + (H% 2), where% indicates the modulus operation.
3. The method as claimed in claim 1, wherein the side length of the square small block in 2) is selected from three numbers of 8, 4 and 2: if in fig. IbThe width and the height of the square can be completely divided by 8, and the side length of the small square block is set to be 8; if in fig. IbThe width and the height of the square can not be divided by 8 but can be divided by 4, and the side length of the small square block is set to be 4; if in fig. IbThe width and the height of the square can not be divided by 4 but can be divided by 2, and the side length of the small square block is set to be 2;
4. the method of claim 1, wherein the mean sequence A in 2) is represented as follows:
A＝{a1,a2,…,ai,…,am}
wherein a isiMeans representing the ith region of interest block, i ∈ [1, m]And m is the number of region of interest blocks.
5. The method of claim 1, wherein the gray scale cloud mask N in 3c)bA set omega of n non-zero gray values, where the gray value is kkForm the k-th cloud region, omegak＝{Nb(i,j)|Nb(i,j)＝k}，k∈[1,n]And n is the number of the cloud area types.
6. The method according to claim 1, wherein the dictionary D in 4) can be expressed as:
D＝{d1,d2,…,di,…,dn}
wherein d isiRepresenting the mean value of the edge pixels of the ith cloud area, i belongs to [1, n ∈]When filling, the cloud-containing remote sensing image I after expansionbAll the pixel points in the ith cloud area are filled with di。
7. The method of claim 1, wherein the compression coding in 7) employs a JPEG2000 or JPEG-LS compression algorithm.
8. The method of claim 1, wherein 8) the receiving end decodes the received code stream S by using a decoding algorithm matching with the encoding of the transmitting end.
9. The method of claim 7, wherein the compression encoding is performed using the JPEG2000 algorithm, which in turn comprises the preprocessing of the image IpAnd performing the processes of component transformation, fragmentation processing, wavelet transformation, quantization, entropy coding, code rate control and code stream organization.
10. Method according to claim 7, characterized in that the compression coding is performed using the JPEG-LS algorithm, which in turn comprises the pre-processing of the image IpAnd performing processes of context modeling, prediction, residual calculation, parameter updating and Golomb-Rice coding.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011350894.9A CN112465846B (en) | 2020-11-26 | 2020-11-26 | Cloud-containing remote sensing image compression method based on filling strategy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011350894.9A CN112465846B (en) | 2020-11-26 | 2020-11-26 | Cloud-containing remote sensing image compression method based on filling strategy |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112465846A true CN112465846A (en) | 2021-03-09 |
| CN112465846B CN112465846B (en) | 2023-04-07 |
Family
ID=74808011
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011350894.9A Active CN112465846B (en) | 2020-11-26 | 2020-11-26 | Cloud-containing remote sensing image compression method based on filling strategy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112465846B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113470127A (en) * | 2021-09-06 | 2021-10-01 | 成都国星宇航科技有限公司 | Optical image effective compression method based on satellite-borne cloud detection |
| CN114022790A (en) * | 2022-01-10 | 2022-02-08 | 成都国星宇航科技有限公司 | Cloud layer detection and image compression method and device in remote sensing image and storage medium |
| CN114845042A (en) * | 2022-03-14 | 2022-08-02 | 南京大学 | Camera automatic focusing method based on image information entropy |
| CN115829838A (en) * | 2022-11-23 | 2023-03-21 | 爱芯元智半导体(上海)有限公司 | Image edge expanding circuit, chip and method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093243A (en) * | 2013-01-24 | 2013-05-08 | 哈尔滨工业大学 | High resolution panchromatic remote sensing image cloud discriminating method |
| JP5396559B1 (en) * | 2013-04-17 | 2014-01-22 | 学校法人 埼玉医科大学 | Image encoding apparatus, image encoding method, image encoding program, and recording medium |
| WO2015075185A2 (en) * | 2013-11-21 | 2015-05-28 | Centre National D'etudes Spatiales | Method and device for digital image compression |
| US20170161584A1 (en) * | 2015-12-07 | 2017-06-08 | The Climate Corporation | Cloud detection on remote sensing imagery |
| CN110555818A (en) * | 2019-09-09 | 2019-12-10 | 中国科学院遥感与数字地球研究所 | method and device for repairing cloud region of satellite image sequence |
-
2020
- 2020-11-26 CN CN202011350894.9A patent/CN112465846B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093243A (en) * | 2013-01-24 | 2013-05-08 | 哈尔滨工业大学 | High resolution panchromatic remote sensing image cloud discriminating method |
| JP5396559B1 (en) * | 2013-04-17 | 2014-01-22 | 学校法人 埼玉医科大学 | Image encoding apparatus, image encoding method, image encoding program, and recording medium |
| WO2015075185A2 (en) * | 2013-11-21 | 2015-05-28 | Centre National D'etudes Spatiales | Method and device for digital image compression |
| US20170161584A1 (en) * | 2015-12-07 | 2017-06-08 | The Climate Corporation | Cloud detection on remote sensing imagery |
| CN110555818A (en) * | 2019-09-09 | 2019-12-10 | 中国科学院遥感与数字地球研究所 | method and device for repairing cloud region of satellite image sequence |
Non-Patent Citations (3)
| Title |
|---|
| HUAICHAO WANG ET AL: "An Invalid Cloud Region Masking Method for Remote Sensing Image Compression", 《PATTERN RECOGNITION AND IMAGE ANALYSIS》 * |
| 王静等: "含无效云区域遥感图像在轨压缩方法", 《中国空间科学技术》 * |
| 赵威: "遥感云图感兴趣区域压缩算法研究", 《中国优秀硕士学位论文全文数据库》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113470127A (en) * | 2021-09-06 | 2021-10-01 | 成都国星宇航科技有限公司 | Optical image effective compression method based on satellite-borne cloud detection |
| CN114022790A (en) * | 2022-01-10 | 2022-02-08 | 成都国星宇航科技有限公司 | Cloud layer detection and image compression method and device in remote sensing image and storage medium |
| CN114845042A (en) * | 2022-03-14 | 2022-08-02 | 南京大学 | Camera automatic focusing method based on image information entropy |
| CN114845042B (en) * | 2022-03-14 | 2023-03-10 | 南京大学 | Camera automatic focusing method based on image information entropy |
| CN115829838A (en) * | 2022-11-23 | 2023-03-21 | 爱芯元智半导体(上海)有限公司 | Image edge expanding circuit, chip and method |
| CN115829838B (en) * | 2022-11-23 | 2023-08-11 | 爱芯元智半导体(上海)有限公司 | Image edge-expanding circuit, chip and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112465846B (en) | 2023-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112465846B (en) | Cloud-containing remote sensing image compression method based on filling strategy | |
| CN108028941B (en) | Method and apparatus for encoding and decoding digital images by superpixel | |
| EP1591963B1 (en) | Adaptive quantisation of a depth map | |
| KR100974177B1 (en) | Method and apparatus for using random field models to improve picture and video compression and frame rate up conversion | |
| CN111866521B (en) | Video image compression artifact removing method | |
| CN1157080A (en) | Device and method for coding video pictures | |
| TW201840185A (en) | Non-local adaptive loop filter processing | |
| CN112399176B (en) | Video coding method and device, computer equipment and storage medium | |
| CN108810534B (en) | Image compression method based on direction lifting wavelet and improved SPIHT under Internet of things | |
| CN114842216A (en) | Indoor RGB-D image semantic segmentation method based on wavelet transformation | |
| CN115941943A (en) | HEVC video coding method | |
| US6587588B1 (en) | Progressive image decoder for wavelet encoded images in compressed files and method of operation | |
| JP4449400B2 (en) | Image encoding apparatus and method, program, and recording medium | |
| CN113947538A (en) | Multi-scale efficient convolution self-attention single image rain removing method | |
| JP2007325262A (en) | Method of video compression | |
| Alsayyh et al. | A Novel Fused Image Compression Technique Using DFT, DWT, and DCT. | |
| US8428381B2 (en) | Image compression method with variable quantization parameter | |
| WO2020237759A1 (en) | System and method for effectively representing and coding image | |
| CN112565756B (en) | Cloud-containing remote sensing image compression method based on quantization strategy | |
| KR100381204B1 (en) | The encoding and decoding method for a colored freeze frame | |
| CN111107360B (en) | Spectrum-space dimension combined hyperspectral image lossless compression method and system | |
| CN114386595A (en) | SAR image compression method based on super-prior-check architecture | |
| CN114463449A (en) | Hyperspectral image compression method based on edge guide | |
| Arockia Raj et al. | Turtle edge encoding and flood fill based image compression scheme | |
| CN1396769A (en) | Compression method and system for moving image information |
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 |