CN102495405A - Evaluation method of TM/ETM (thematic mapper/enhanced thematic mapper) and image-based atmospheric correction product quality - Google Patents
Evaluation method of TM/ETM (thematic mapper/enhanced thematic mapper) and image-based atmospheric correction product quality Download PDFInfo
- Publication number
- CN102495405A CN102495405A CN2011103885408A CN201110388540A CN102495405A CN 102495405 A CN102495405 A CN 102495405A CN 2011103885408 A CN2011103885408 A CN 2011103885408A CN 201110388540 A CN201110388540 A CN 201110388540A CN 102495405 A CN102495405 A CN 102495405A
- Authority
- CN
- China
- Prior art keywords
- image
- sample
- product
- etm
- atmospheric correction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention relates to an evaluation method of quality of a TM/ETM (thematic mapper/enhanced thematic mapper) and image-based atmospheric correction product. The method comprises the following steps of: 1, acquiring a product image, and randomly selecting a reference image; 2, matching a geographic coordinate system between the product image and the reference image in the step 1; 3,selecting a plurality of samples from the product image and the reference image by adopting a systematic sampling method, and respectively calculating spectral values of corresponding samples of the product image and reference image on the reference image and the product image; 4, respectively identifying constant land feature samples of the product image and the reference image in the step 2, and saving an effective constant land feature sample; and 5, analyzing the atmospheric correction quality of the product image according to the effective constant land feature sample in the step 4. The method has the advantages that: important basic image quality information can be provided to regional or global earth surface change researches, and PIFs (physical interfaces) samples can be accurately identified on the TM/ETM and image with different time phases/season phases.
Description
Technical field
The present invention relates to a kind of product quality evaluation method, especially relate to a kind of TM/ETM+ image atmospheric correction product quality evaluation method.
Background technology
The TM/ETM+ image that Landsat (Landsat) obtains is owing to advantages such as its moderate resolution, long-term earth observations continuously, and the important foundation data that become the research whole world/regional change are come, and in numerous areas, have obtained large-scale application.Yet; The steam that comprises in the atmosphere, ozone and gasoloid to function influences such as the absorption of solar radiation and scatterings the availability of remote sensing images; Usually need to adopt physics radiation mode or statistical method to remove atmospheric effect to obtain face of land real reflectance; Gas is proofreaied and correct calibration model, the atmospheric environmental parameters used and has been comprised many uncertain factors; Also comprise determinacy scarcely in the surface reflectivity product that causes inverting to obtain, and influenced face of land parameter extraction precision such as vegetation index, leaf area index.Adopting scientific approach that the atmospheric correction product quality of TM/ETM+ is estimated, be not only the necessary ways of improving production quality, also is to improve pressing for of product application result reliability.
At present both at home and abroad the basic ideas of relevant remote sensing image atmospheric correction quality control all are to adopt statistical method, through the spectral value and the quality of the consistance between " truly " spectral value with the evaluation product quality of some ground table object on the comparative analysis product image.Different according to " true value " spectrum source, detection method for quality mainly can be divided into following 2 types.(1) ground synchronous observation proof method.Promptly in the test block, select some homogeneities, smooth zone as the checking sample, when remote sensing satellite passes by, utilize terrestrial radiation surveying instrument simultaneous observation sample earth surface reflection rate, it is evaluated the atmospheric correction product quality as " true value ".(2) with reference to the image contrast method.As " true value " source, choose some ground object sample with high-quality surface reflectivity image, comparative sample is in product image and corresponding spectral value consistance with reference to image, and then the evaluation product quality.
In the above-mentioned 2 series products verification methods, the former " true value " data reliability is good, but cost is higher, and the length that expends time in is not suitable on the whole world/regional scale, carrying out large-scale application, also can't verify history image atmospheric correction result; The evaluation result that the latter obtains then is a kind of relative mass, has only when good with reference to the adjustment of image quality and just can obtain evaluation result comparatively reliably when having close spectral band, identical observation time with the product image.And the latter requires to use the high-quality image image as a reference that has an identical observation time with the product image usually; This method can not be to carrying out quality assessment at the TM/ETM+ image atmospheric correction product that obtains before this, and MODIS image and TM/ETM+ image to a certain degree also can be to the quality evaluation result deleterious impact in the difference of aspects such as spatial resolution, projected coordinate system.
Summary of the invention
The present invention mainly is that to solve the existing in prior technology cost higher, and the length that expends time in is not suitable on the whole world/regional scale, carrying out large-scale application, the technical matters that also can't verify etc. history image atmospheric correction result; It is with reference to image that a kind of global surface reflectivity product with Landsat is provided; Can realize the consistance of TM/ETM+ audio and video products quality of many phases of history is estimated; Can crucial base image quality information be provided for the regional or global face of land changes research, and then guarantee a kind of TM/ETM+ image atmospheric correction product quality evaluation method of result of study precision.
The present invention also have a purpose be solve existing in prior technology can not be to carrying out quality assessment at the TM/ETM+ image atmospheric correction product that obtains before this, and MODIS image and TM/ETM+ image to a certain degree also can be to the technical matterss of quality evaluation result deleterious impact etc. in the difference of aspects such as spatial resolution, projected coordinate system; A kind of PIFs sample automatic identifying method of estimating towards product quality of having proposed is provided, can have had not on the TM/ETM+ image of phase/aspect simultaneously a kind of TM/ETM+ image atmospheric correction product quality evaluation method that accurate recognition comparatively goes out the PIFs sample.
Above-mentioned technical matters of the present invention mainly is able to solve through following technical proposals:
A kind of TM/ETM+ image atmospheric correction product quality evaluation method is characterized in that, may further comprise the steps:
Step 2, coupling in the step 1 the product image and with reference to the geographic coordinate system between the image;
Step 3; Adopt the method for systematic sampling; Respectively according to face of land distribution situation from the product image with reference to extracting some samples the image; And respectively the counting yield image and with reference to the corresponding sample of image with reference to the spectral value on image and the product image, wherein each band spectrum value of sample is sample is gathered spectral value at this wave band pixel a mean value;
Step 4, carry out respectively in the above-mentioned steps 2 the product image and with reference to the corresponding sample of image at the product image with reference to the constant ground object sample identification of image, and preserve effective constant ground object sample;
Step 5 adopts the atmospheric correction quality of regression analysis according to effective constant atural object sample analysis product image in the step 4.
In above-mentioned a kind of TM/ETM+ image atmospheric correction product quality evaluation method, in the described step 1, be through the high-quality image of checking on the spot or the surface reflectivity product of U.S. NASA download with reference to image.
In above-mentioned a kind of TM/ETM+ image atmospheric correction product quality evaluation method; In the described step 2; This coupling step comprises comparative product image and whether consistent with reference to projective parameter, the ellipsoidal parameter of the spatial resolution between the image, geographic coordinate system; If inconsistent, then carry out the consistance of parameter and handle back execution in step 3.
In above-mentioned a kind of TM/ETM+ image atmospheric correction product quality evaluation method; In the described step 3; The concrete grammar of dividing some samples is: according to the face of land distribution situation of product image; Every the sample of a 3*3 or 5*5 pixel size is set, is used to extract and comparative product image and with reference to the spectrum consistance between the image at a distance from fixing ranks; And cover complexity according to the face of land of test block and select to carry out:
When the face of land of test block covered complicacy, constant atural object selects sampling interval to be: whenever sample at a distance from 10 pixels, sample size was: 3*3;
When the face of land, test block covers singlely, and constant atural object is when more, and constant atural object selects sampling interval to be: whenever sample at a distance from 50-200 pixel, sample size is: 5*5.
In above-mentioned a kind of TM/ETM+ image atmospheric correction product quality evaluation method, the calculating of spectral value is to calculate some samples of above-mentioned steps 3 divisions at TM/ETM+ image wave band 1,2 in the described step 3; 3; The pixel spectrum mean value and the coefficient of variation on 4,5,7.
In above-mentioned a kind of TM/ETM+ image atmospheric correction product quality evaluation method, in the described step 4, constant ground object sample identification may further comprise the steps:
Step 4.1, sample homogeney determining step: establish the pixel set that sample S comprised and be C on wave band i, then sample S is the pixel spectrum mean value of C at the spectral value of this wave band; The homogeney that the coefficient of variation of sample on wave band i is used to measure sample, its value is pixel spectral value standard deviation and the ratio of mean value of set C, when sample when each band spectrum coefficient of variation satisfies following condition simultaneously on reference to image and product image; This sample is thought the homogeneity sample, and rejects invalid non-homogeneous sample: the coefficient of variation≤0.15 of sample on the wave band 1 of TM/ETM+ image, at wave band 2; 3; The coefficient of variation on 4,5,7≤0.1;
Step 4.2, the homogeneity sample of having accomplished to step 4.1 carries out the ground object sample that continues to have after the NDVI index is judged, rejects invalid non-constant ground object sample, and concrete steps are:
Step 4.21 is calculated the homogeneity sample respectively with reference to the normalized differential vegetation index NDVI on image and the product image;
Step 4.22 is calculated the coefficient R of homogeneity sample each band spectrum value on reference to image and product image and the absolute value delta NDVI of normalized differential vegetation index difference;
Step 4.23, when R>=0.9, and Δ NDVI≤0.1 o'clock, this homogeneity sample is considered to effective constant ground object sample, rejects invalid sample simultaneously.
In above-mentioned a kind of TM/ETM+ image atmospheric correction product quality evaluation method, in the described step 5, the concrete determining step of atmospheric correction quality is following:
formula three
In the formula: the set of X each band spectrum value that is all samples on reference to image, the set of all samples of Y corresponding spectral value on the product image, X is the mean value of X, and Y is the mean value of Y, and N is that sample size and image wave band number are long-pending;
Step 2 is estimated TM/ETM+ image atmospheric correction product quality: when the value of a and R approaches 1; And b and RMSD approached 0 o'clock, showed the product image and with reference to spectrum high conformity between the image, and image atmospheric correction quality is high.
Therefore; The present invention has following advantage: 1. the global surface reflectivity product with Landsat is with reference to image; Can realize the consistance of TM/ETM+ audio and video products quality of many phases of history is estimated; Can crucial base image quality information be provided for the regional or global face of land changes research, and then guarantee the result of study precision; 2. the PIFs sample automatic identifying method estimated towards product quality has been proposed, can accurate recognition goes out the PIFs sample having not on the TM/ETM+ image of phase/aspect simultaneously comparatively.
Description of drawings
Fig. 1 is a TM/ETM+ image atmospheric correction quality control method flow diagram.
Fig. 2 is a product image spectrum sample method.
Fig. 3 is the spectral signature explanation of constant ground object sample.
Fig. 4 TM/ETM+ image atmospheric correction product quality is estimated synoptic diagram.
Fig. 5 a the present invention tests in the case, and test block P124R039 (30.30 ° of N of center latitude, middle longitude centroid are 112.22 ° of E) is at the synoptic diagram of observation on September 10th, 1999 without image " P124R039_19990910_TOA " quality evaluation result of atmospheric correction.
Fig. 5 b the present invention tests in the case; Test block P124R039 (30.30 ° of N of center latitude, middle longitude centroid are 112.22 ° of E) adopts the FLAASH model to carry out the synoptic diagram of image " P124R039_19990910_FLA " quality evaluation result of atmospheric correction in observation on September 10th, 1999.
Fig. 5 c the present invention tests in the case, and test block P124R039 (30.30 ° of N of center latitude, middle longitude centroid are 112.22 ° of E) is at the synoptic diagram of observation on January 5th, 2002 without image " P124R039_20020105_TOA " quality evaluation result of atmospheric correction.
Fig. 5 d the present invention tests in the case; Test block P124R039 (30.30 ° of N of center latitude, middle longitude centroid are 112.22 ° of E) adopts the FLAASH model to carry out the synoptic diagram of image " P124R039_20020105_FLA " quality evaluation result of atmospheric correction in observation on January 5th, 2002.
Fig. 5 e the present invention tests in the case, and test block P129R031 (41.76 ° of N of center latitude, 107.85 ° of E of middle longitude centroid) is at the synoptic diagram of observation on May 3rd, 2003 without image " P129R031_20030503_TOA " quality evaluation result of atmospheric correction.
Fig. 5 f the present invention tests in the case; Test block P129R031 (41.76 ° of N of center latitude, 107.85 ° of E of middle longitude centroid) adopts the FLAASH model to carry out the synoptic diagram of image " P129R031_20030503_FLA " quality evaluation result of atmospheric correction in observation on May 3rd, 2003.
Fig. 5 g the present invention tests in the case, and test block P129R031 (41.76 ° of N of center latitude, 107.85 ° of E of middle longitude centroid) is at the synoptic diagram of observation on September 8th, 2009 without image " P129R031_20090908_TOA " quality evaluation result of atmospheric correction.
Fig. 5 h the present invention tests in the case; Test block P129R031 (41.76 ° of N of center latitude, 107.85 ° of E of middle longitude centroid) adopts the FLAASH model to carry out the synoptic diagram of image " P129R031_20090908_FLA " quality evaluation result of atmospheric correction in observation on September 8th, 2009.
Fig. 6 a the present invention tests in the case, the synoptic diagram of the NDVI exponential distribution statistics of the constant ground object sample of gathering on 4 width of cloth pictures of test block P124R039 (30.30 ° of N of center latitude, middle longitude centroid are 112.22 ° of E).
Fig. 6 b the present invention tests in the case, the synoptic diagram of the NDVI exponential distribution statistics of the constant ground object sample of gathering on 4 width of cloth pictures of test block P129R031 (41.76 ° of N of center latitude, 107.85 ° of E of middle longitude centroid).
Embodiment
Pass through embodiment below, and combine accompanying drawing, do further bright specifically technical scheme of the present invention.
Embodiment:
At first introduce the related theoretical foundation of present embodiment:
Owing to receive the influence of factors such as sun altitude, solar distance, calendar variation and atmospheric effect, same atural object (particularly vegetation) exists than big-difference at the image spectral reflectivity that different time observation obtains usually.Therefore, the time mutually different remote sensing images can not directly compare usually, (Pseudo Invariant Features, PIFs), the spectrum consistance through PIFs relatively is the consistance of image spectrum relatively indirectly need on image, to select pseudo-invariant features point.Pseudo-invariant features point is one type of atural object that spectral signature does not change in time, like desert, artificial covering etc.This characteristic of PIFs is normally used for not the normalization of phase remote sensing image simultaneously and handles, and also is the basis that does not compare between the phase TM/ETM+ image simultaneously among the present invention.
The big pneumatic jack reflectivity of remote sensing image (the Top of Atmospheric that obtains through operations such as radiation calibrations; TOA) eliminated the influence that is brought by sun altitude, solar distance difference, atmospheric correction is then to a certain degree eliminated the influence that atmospheric conditions difference causes.In theory, should be basic identical at the PIFs spectral reflectivity that different time observation obtains, if exist than big-difference, then main cause is the incorrect atmospheric effect of eliminating.Based on above-mentioned principle; Design TM/ETM+ image atmospheric correction product quality evaluation method comprises 3 committed steps: 1. select high-quality TM/ETM+ surface reflectivity product image as a reference; And according to suitable sample size and the SI of image face of land covering characteristics design; To be benchmark, adopt systemic sampling method to generate a series of object spectrum samples with reference to image; 2. each sample of comparative analysis is at the product image with reference to the spectral value on the image, and therefrom extracts the PIFs sample; 3. obtain the PIFs sample respectively and carrying out consistance with reference to the spectral value on image and the product image relatively, estimate product image relative mass.The concrete grammar flow process is seen Fig. 1.
The product image is to obtain abundant PIFs sample with the purpose of carrying out spectrum sample with reference to image is used for quality assessment.Though still there is pixel left and right sides site error half in the TM/ETM+ image that different time obtains through geometric accurate correction between the image, and since the existence of face of land proximity effect be not suitable for directly carrying out the comparison between the image by pixel.For obtaining abundant PIFs sample, spectrum sample mainly adopts systemic sampling method, and the basic norm of following the PIFs Sample selection designs, that is: 1. PIFs homogeneity as far as possible; 2. PIFs should have than large tracts of land to reduce the influence of face of land proximity effect; 3. the PIFs spectral value has distribution in a big way.
Based on mentioned above principle, the key of image spectrum sample conceptual design is to confirm sample size and SI: the sample-sized design is excessive, then is difficult to find the sample of homogeneity, too smallly then is unfavorable for eliminating proximity effect; SI is big more, and then counting yield is fast more, and the sample size of acquisition is less relatively, otherwise then can obtain more sample.Sample size and SI mainly are provided with according to face of land coverage condition in the test block: face of land cover type is intricate in the test block, and PIFs atural object selects less sample and SI to enlarge the PIFs hunting zone more after a little while; Single relatively and then can select bigger sample and SI to improve counting yield when having large stretch of PIFs atural object when cover type in the test block.Following table be one according to test of many times draw can be for reference image spectrum sample scheme parameter configuration.
| Image face of land Cover Characteristics | Sample size (pixel) | SI (pixel) |
| Face of land cover type is intricate, and constant atural object is less | 3*3 | 10-30 |
| Face of land cover type is single, and large stretch of constant atural object is arranged | 5*5 or 7*7 | 50-100 |
The concrete operations step of present embodiment is following:
(1) chooses the reference image of the high-quality TM/ETM+ earth surface reflection rate image of existing first phase process checking in the product test block as the quality assessment of product image atmospheric correction.Can pass through the high-quality image of checking on the spot with reference to image, also can be from the surface reflectivity product of U.S. NASA download;
Need to prove that 1. the high-quality image is commonly referred to be the real spectrum value that spectral value on the image is in close proximity to the face of land." approaching " mainly also is to carry out regretional analysis through image polishing wax value and " real spectrum value " to measure.But there is not unified standard to be used to stipulate what is " high-quality " image as yet at present both at home and abroad; Rule of thumb make judgement mostly with the accuracy requirement of image use; As when regression coefficient during, can think " high-quality " image greater than the numerical value of certain expectation.2. the seminar of US National Aeronautics and Space Administration (NASA), American National geologic examination office (USGS) and Univ Maryland-Coll Park USA has developed the surface reflectivity product (Global Land Survey Surface Reflectance Product) of the TM/ETM+ in the cover global range jointly.(product of MODIS has had much human to carry out research and checking on the spot to this product owing to carried out the consistance evaluation with the MODIS audio and video products; Be considered to reliable, high-quality; All obtained proof in the many papers of this conclusion at home and abroad); Evaluation result is comparatively desirable, and all data all provide quality evaluation result information, so most of image (accounting for more than the 80-90% greatly) of this data centralization can be considered to the image of " high-quality ".(4) up to the present, in the world, the image that certain zone can be downloaded from the Global Land Survey Surface Reflectance Product data centralization of NASA has only the 2-3 width of cloth.When carrying out quality assessment, can select the season and the image download as a reference of the immediate image of product image of a shooting.(5) the GLS Surface Reflectance Product image data collection that provides of NASA can be used as the main reference Data Source of this research really.
(2) comparative product image and whether consistent with reference to projective parameter, the ellipsoidal parameter of the spatial resolution between the image, geographic coordinate system, if inconsistent, the consistance that then requires the user at first to carry out parameter is handled, otherwise can't compare; Here, need to prove:
1. what remote sensing image was taken is the surface of ground ball, and the process that converts this sphere on the plane calls " map projection ".
2. owing in the world when earth ellipsoid is carried out modeling, different mathematical projection methods and earth ellipsoid face are arranged.The difference of projecting method can obtain distinct result with the different of earth ellipsoid face parameter; For example the isometric projection method may be produced area generation great variety and angle does not change; And the equal area projection rule can produce angle and deform, and area remains unchanged.
3. remote sensing image can be selected different map projection's methods and earth ellipsoid face parameter according to the needs that use, and employed projective parameter of remote sensing image and ellipsoidal parameter can be read identification by computer program as the part of image metadata usually.
4. whether this method is not handled and is changed geographic coordinate system and ellipsoidal parameter, only need 2 employed geographic coordinate systems of image of comparison consistent with the ellipsoidal parameter setting.When having only the coordinate system parameter value that uses when 2 images in full accord, just can compare; Otherwise will cause the comparative result mistake.Therefore, when carrying out quality assessment, the coordinate parameters that requires the user to unify 2 image datas earlier carries out the operation that subsequent quality is estimated again.
5. the unification of image projecting, ellipsoidal parameter and conversion can be accomplished in a lot of ripe business softwares, like ArcGIS etc., but is not the concerned issue of this method.
(3) adopt the method for systematic sampling,, whenever the sample of a 3*3 or 5*5 pixel size is set, be used to extract and comparative product image and with reference to the spectrum consistance between the image at a distance from fixing ranks according to the face of land distribution situation of product image; When the face of land of test block covers more complicated; Constant atural object (desert, artificial covering etc., cement pavement, airport etc.) is than less sampling interval of more options (whenever sampling at a distance from 10 pixels) and sample size (3*3); When the face of land, test block covers more single; And when constant atural object is more, can select bigger sampling interval (every separated 50-200 pixel sampled) and sample size (5*5); Generally in the plains region, arable land, waters, residential area are a lot, and on space distribution relatively more assorted, chaotic in, just can think relative complex, and for desert, desert, this area thinks that face of land cover type is simple relatively.
(4) to obtain the spectrum samples collection through systemic sampling method, at first calculate pixel spectrum mean value and the coefficient of variation of sample on TM/ETM+ image wave band 1,2,3,4,5,7 (coefficient of variation, C.V);
(5) as shown in Figure 2, establish the pixel set that sample S comprised and be C on wave band i, then sample S is the pixel spectrum mean value CAVG of C at the spectral value SBandi of this wave band; The homogeney that the coefficient of variation CCV of sample on wave band i is used to measure sample, its value is the pixel spectral value standard deviation C σ of set C and the ratio of mean value CAVG.
(6) when sample when each band spectrum coefficient of variation satisfies following condition simultaneously on reference to image and product image; This sample is thought the homogeneity sample, and rejects invalid non-homogeneous sample: the coefficient of variation≤0.15 of sample on the wave band 1 of TM/ETM+ image, at wave band 2; 3; The coefficient of variation on 4,5,7≤0.1;
(7) characteristic as shown in Figure 3 according to constant ground object sample designs its automatic distinguishing indexes and corresponding threshold interval.When certain sample met the following conditions simultaneously, this sample was identified as the PIF sample: 1. R is SPRO and SREF related coefficient, has only that this sample just possibly be the PIF point when R>=0.9; 2. Δ NDVI≤0.1, Δ NDVI is normalized differential vegetation index (Normalized Difference Vegetation Index, the NDVI) absolute value of difference of sample on 2 phase images.Since surface vegetation be image with the main type of ground objects that changes season, in order to discern the PIFs sample more accurately, on the basis of 2 phase of analyzing samples correction of image property (R), consider that simultaneously sample NDVI index variation can further remove non-PIFs sample;
(8) reject invalid non-PIF sample;
(9) adopt the linear regression model (LRM) analysis to analyze the PIFs sample, obtain the regression equation of formula (1), and calculate coefficient R (formula 2) and sample spectrum root-mean-square error RMSD (formula 3) at the product image with reference to the spectrum consistance on the image:
(1)
(2)
In the formula: the set of X each band spectrum value that is all samples on reference to image, the set of all samples of Y corresponding spectral value on the product image, X is the mean value of X, and Y is the mean value of Y, and N is that sample size and image wave band number are long-pending.
(10) estimate TM/ETM+ image atmospheric correction product quality: when the value of a and R approaches 1, when b and RMSD approach 0, show product image and better with reference to spectrum consistance between the image, image atmospheric correction quality is higher; Otherwise product adjustment of image the possibility of result existing problems have much room for improvement.
Be concrete example below:
(1) Experimental Area of experiment case is described: the present invention tests case and chooses 2 zones with different surface Cover Characteristics as the test block.Wherein: 1. the position of test block 1 in the WRS-2 of Landsat (Worldwide Reference System) coordinate system is Path=124, Row=039, and 30.30 ° of N of center latitude, middle longitude centroid are 112.22 ° of E.This test block mainly is positioned at the Jianghan Plain in Hubei, and there is a small amount of mountain area in western part, and face of land cover type is mainly ploughs and forest, and constant atural object is less, face of land cover type relative complex; 2. the WRS-2 coordinate position of test block 2 is Path=129, Row=031,41.76 ° of N of center latitude, 107.85 ° of E of middle longitude centroid.This test block mainly is positioned at the Baya ur, NeiMengGu area, and face of land cover type is mainly desert/desertificated area, and the southeast has a small amount of arable land and meadow, and face of land cover type is single relatively.
(2) the reference image of experiment case use: the high-quality that experiment is used obtains from the Global Land Survey Surface reflectance of University of Maryland data centralization with reference to image.Essential information sees the following form.
| Image name | Sensor | The acquisition time |
| P124R039_20010915_GLS | ETM+ | 2001.9.15 |
| P129R031_20070903_GLS | ETM+ | 2007.9.3 |
Wherein: 1. the acquisition time with reference to image P124R039_20010915_GLS is September 15 calendar year 2001; Red part is sheet arable land, artificial covering and the river shoal sand ground etc. that Jianghan Plain has just gathered on the image; Green is the forest land that is distributed in western mountainous areas, and water body is blue; 2. the acquisition time with reference to image P129R031_20070903_GLS is on September 3rd, 2007, and southeastern vegetation is in the season of growth on the image, is rendered as green, and remainder is mainly desert and desertification area.
(3) the product image of experiment use: the image of each test block free download 2 scape Various Seasonal from the Landsat L1T product of U.S. NASA is as experiment product image; Wherein 1 scape is with identical season with reference to image; 1 scape is with different season with reference to image, and experiment image essential information is seen table 5.Every scape image all utilizes its metadata that carries in ENVI software, successively to accomplish: 1. radiation calibration is scaled to atmospheric envelope top reflectivity (TOA) with the digital value (DN) of raw video; 2. carry out atmospheric correction with the Flaash module of ENVI, with atmospheric envelope top reflectivity correction be surface reflectivity (Surface Reflectance, SR).Be convenient and calculate that the numerical value of atmospheric envelope top reflectivity image and surface reflectivity image all multiply by 10000 times, pixel numerical value interval is 0-10000, and is consistent with reference to image.The image essential information of examine sees the following form:
Annotate: in the image name, with " TOA " image of ending be the image without atmospheric correction, FLA has carried out the image behind the atmospheric correction for the FLAASH module of use ENVI.
(4) write computer program, the constant atural object method of sampling that designs in the realization present embodiment, and automatically identify constant ground object sample.Wherein: the face of land cover type of test block 1 mainly is to plough and forest, and constant atural object is less, and for obtaining more PIFs sample, 10 pixels in every interval generate the spectrum samples that size is 3*3; The face of land cover type of test block 2 is mainly desert/desert, and constant atural object is more, so adopt 50 pixels in every interval to generate the sampling plan that size is the sample of 5*5.
(5) estimate the atmospheric correction quality of each image, the result sees the following form, relevant statistical distribution map accompanying drawing 5.In the accompanying drawing 5, indicated the product image name on the longitudinal axis of coordinate system, transverse axis has indicated the corresponding reference image name.
| Image name | The PIF number of samples | Quality evaluation result |
| P124R039_19990910_TOA | 2988 | Y=0.65X+482.05,? R=0.88,? RMSD=496.70 |
| P124R039_19990910_FLA | 2844 | Y=0.89X+3.07,? R=0.97,? RMSD=285.84 |
| P124R039_20020105_TOA | 62 | Y=0.4739X+956.57,? R=0.81,? RMSD=597.75 |
| P124R039_20020105_FLA | 52 | Y=0.92X+374.73,? R=0.90,? RMSD=445.66 |
| P129R031_20090908_TOA | 8829 | Y=0.62X+743.41,? R=0.97,? RMSD=594.70 |
| P129R031_20090908_FLA | 8395 | Y=0.89X+85.26,? R=0.9893,? RMSD=313.63 |
| P129R031_20030503_TOA | 7973 | Y=0.6339X+1040.12,? R=0.97,? RMSD=484.27 |
| P129R031_20030503_FLA | 8267 | Y=0.86X+518.34,? R=0.98,? RMSD=303.62 |
(6) can know from sample distribution and sample correlations analysis result; The quality of image through behind the atmospheric correction obviously improves; Be mainly reflected in: through the image of atmospheric correction; Its PIFs sample spectral value regression coefficient and related coefficient approach 1 more with respect to the image without the mistake atmospheric correction, and the root-mean-square error RMSD of sample also reduces relatively.This shows that the evaluation result that is drawn by invention proposition TM/ETM+ image atmospheric correction product quality evaluation model conforms to TM/ETM+ image atmospheric correction quality situation basically.
(7) accuracy of PIFs sample extraction directly influences the reliability of evaluation result, for further verifying the rationality of the quality evaluating method that the present invention proposes, must estimate and analyze from 2 aspects of quality and quantity the PIFs sample that each image extracts.
In quantity, each is tested the PIFs sample that image extracts and has following characteristics: the sample size that extracts on 1 image of (1) test block obviously is less than the sample number that extracts on 2 images of test block; (2) totally approach the sample number that extracts without on the atmospheric correction image through the sample number that extracts on the image behind the atmospheric correction; (3) test block 1 with reference to the sample size that extracts on the identical aspect of image much larger than with reference to the sample size that extracts on the different image of image aspect; And on test block 2, the latter then is slightly less than the former.The PIFs sample size characteristics of extracting on each image have reflected that basically the table of 2 test blocks covers characteristics; That is: to be capped area bigger in test block 1 interplantation; Constant atural object distributed areas are less; So the sample size that extracts is less, product image and the sample number that does not extract simultaneously with reference to the image aspect are still less; It is less that test block 2 interplantations are capped area, and the distribution in the constant atural object-desert of large tracts of land is arranged, and it is more to extract sample size, and the sample size that extracts on the different aspect image is also approaching basically.
The main statistical sample of PIFs sample quality analysis is concentrated the quantity of vegetation sample.Because the curve of spectrum of surface vegetation areal coverage can be with great change takes place season, therefore, the vegetation number of samples should be the least possible in the PIFs sample.Be vegetation sample size situation in the statistics PIF sample, respectively PIFs sample NDVI exponential distribution situation on test block 1 and the test block 2 product images added up that the result sees Fig. 8.
(8) sample NDVI index on the image is analyzed, when the NDVI index greater than 0.45 the time, this sample promptly is classified as the vegetation sample.In 2 test blocks, can get on the NDVI index statistical Butut of each sample set: in (1) test block 1 with the PIF sample that extracts with reference to the identical image of image aspect, it is the vegetation sample that a half-sample is arranged approximately; With with reference in the PIF sample that extracts on the different image of image aspect, the vegetation sample size is then less than 10%; (2) the PIF sample that extracts on 2 interior 2 the different aspect images of test block almost all is non-vegetation sample.In the test block 1, be September 15 calendar year 2001 with reference to the image observation time, the experiment image identical with its aspect observed on September 10th, 1999, and 2 observation times are very approaching, and it is constant basically that the spectrum of vegetation can be thought.Therefore, it is rational from the sample of P124R039_19990910_TOA and P124R039_19990910_FLA extraction, comprising more vegetation sample.
Specific embodiment described herein only is that the present invention's spirit is illustrated.Person of ordinary skill in the field of the present invention can make various modifications or replenishes or adopt similar mode to substitute described specific embodiment, but can't depart from spirit of the present invention or surmount the defined scope of appended claims.
Claims (7)
1. a TM/ETM+ image atmospheric correction product quality evaluation method is characterized in that, may further comprise the steps:
Step 1 is obtained the product image, and the high-quality TM/ETM+ earth surface reflection rate image of picked at random first phase process checking is as the reference image of product image atmospheric correction quality assessment;
Step 2, coupling in the step 1 the product image and with reference to the geographic coordinate system between the image;
Step 3; Adopt the method for systematic sampling; Respectively according to face of land distribution situation from the product image with reference to extracting some samples the image; And respectively the counting yield image and with reference to the corresponding sample of image with reference to the spectral value on image and the product image, wherein each band spectrum value of sample is sample is gathered spectral value at this wave band pixel a mean value;
Step 4, carry out respectively in the above-mentioned steps 2 the product image and with reference to the corresponding sample of image at the product image with reference to the constant ground object sample identification of image, and preserve effective constant ground object sample;
Step 5 adopts the atmospheric correction quality of regression analysis according to effective constant atural object sample analysis product image in the step 4.
2. a kind of TM/ETM+ image atmospheric correction product quality evaluation method according to claim 1 is characterized in that, in the described step 1, is through the high-quality image of checking on the spot or the surface reflectivity product of U.S. NASA download with reference to image.
3. a kind of TM/ETM+ image atmospheric correction product quality evaluation method according to claim 1; It is characterized in that; In the described step 2; This coupling step comprises comparative product image and whether consistent with reference to projective parameter, the ellipsoidal parameter of the spatial resolution between the image, geographic coordinate system, if inconsistent, then carry out the consistance of parameter and handles back execution in step 3.
4. a kind of TM/ETM+ image atmospheric correction product quality evaluation method according to claim 1; It is characterized in that; In the described step 3; The concrete grammar of dividing some samples is: according to the face of land distribution situation of product image, every at a distance from fixing ranks the sample of a 3*3 or 5*5 pixel size is set, is used to extract and comparative product image and with reference to the spectrum consistance between the image; And cover complexity according to the face of land of test block and select to carry out:
When the face of land of test block covered complicacy, constant atural object selects sampling interval to be: whenever sample at a distance from 10 pixels, sample size was: 3*3;
When the face of land, test block covers singlely, and constant atural object is when more, and constant atural object selects sampling interval to be: whenever sample at a distance from 50-200 pixel, sample size is: 5*5.
5. a kind of TM/ETM+ image atmospheric correction product quality evaluation method according to claim 1; It is characterized in that the calculating of spectral value is to calculate some samples of above-mentioned steps 3 divisions at TM/ETM+ image wave band 1,2 in the described step 3; 3; The pixel spectrum mean value and the coefficient of variation on 4,5,7.
6. a kind of TM/ETM+ image atmospheric correction product quality evaluation method according to claim 1 is characterized in that, in the described step 4, constant ground object sample identification may further comprise the steps:
Step 4.1, sample homogeney determining step: establish the pixel set that sample S comprised and be C on wave band i, then sample S is the pixel spectrum mean value of C at the spectral value of this wave band; The homogeney that the coefficient of variation of sample on wave band i is used to measure sample, its value is pixel spectral value standard deviation and the ratio of mean value of set C, when sample when each band spectrum coefficient of variation satisfies following condition simultaneously on reference to image and product image; This sample is thought the homogeneity sample, and rejects invalid non-homogeneous sample: the coefficient of variation≤0.15 of sample on the wave band 1 of TM/ETM+ image, at wave band 2; 3; The coefficient of variation on 4,5,7≤0.1;
Step 4.2, the homogeneity sample of having accomplished to step 4.1 carries out the ground object sample that continues to have after the NDVI index is judged, rejects invalid non-constant ground object sample, and concrete steps are:
Step 4.21 is calculated the homogeneity sample respectively with reference to the normalized differential vegetation index NDVI on image and the product image;
Step 4.22 is calculated the coefficient R of homogeneity sample each band spectrum value on reference to image and product image and the absolute value delta NDVI of normalized differential vegetation index difference;
Step 4.23, when R>=0.9, and Δ NDVI≤0.1 o'clock, this homogeneity sample is considered to effective constant ground object sample, rejects invalid sample simultaneously.
7. a kind of TM/ETM+ image atmospheric correction product quality evaluation method according to claim 1 is characterized in that in the described step 5, the concrete determining step of atmospheric correction quality is following:
Step 1 adopts linear regression model (LRM) to analyze constant ground object sample at the product image with reference to the spectrum consistance on the image, obtains the regression equation of formula one, and the calculating coefficient R is that formula two is a formula three with sample spectrum root-mean-square error RMSD:
In the formula: the set of X each band spectrum value that is all samples on reference to image, the set of all samples of Y corresponding spectral value on the product image, X is the mean value of X, and Y is the mean value of Y, and N is that sample size and image wave band number are long-pending;
Step 2 is estimated TM/ETM+ image atmospheric correction product quality: when the value of a and R approaches 1; And b and RMSD approached 0 o'clock, showed the product image and with reference to spectrum high conformity between the image, and image atmospheric correction quality is high.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110388540 CN102495405B (en) | 2011-11-30 | 2011-11-30 | Evaluation method of TM/ETM (thematic mapper/enhanced thematic mapper) and image-based atmospheric correction product quality |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110388540 CN102495405B (en) | 2011-11-30 | 2011-11-30 | Evaluation method of TM/ETM (thematic mapper/enhanced thematic mapper) and image-based atmospheric correction product quality |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102495405A true CN102495405A (en) | 2012-06-13 |
| CN102495405B CN102495405B (en) | 2013-06-19 |
Family
ID=46187245
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110388540 Active CN102495405B (en) | 2011-11-30 | 2011-11-30 | Evaluation method of TM/ETM (thematic mapper/enhanced thematic mapper) and image-based atmospheric correction product quality |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102495405B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909260A (en) * | 2017-11-10 | 2018-04-13 | 浙江省地理信息中心 | A kind of natural resource assets are left one's post audit commenting method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5726656A (en) * | 1996-12-19 | 1998-03-10 | Hughes Electronics | Atmospheric correction method for interferometric synthetic array radar systems operating at long range |
| US6484099B1 (en) * | 1999-07-16 | 2002-11-19 | Deutsches Zentrum Fur Luft -Und Raumfahrt E.V. | Process for correcting atmospheric influences in multispectral optical remote sensing data |
| CN1710379A (en) * | 2005-07-05 | 2005-12-21 | 华东师范大学 | Atmosphere correction method of airosol optical thickness of aeronautical high-spectrum remote-sensing inversion boundary layer |
| CN1831555A (en) * | 2005-03-10 | 2006-09-13 | 中国煤炭地质总局航测遥感局 | Processing method for local correction of remote sensing image |
-
2011
- 2011-11-30 CN CN 201110388540 patent/CN102495405B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5726656A (en) * | 1996-12-19 | 1998-03-10 | Hughes Electronics | Atmospheric correction method for interferometric synthetic array radar systems operating at long range |
| US6484099B1 (en) * | 1999-07-16 | 2002-11-19 | Deutsches Zentrum Fur Luft -Und Raumfahrt E.V. | Process for correcting atmospheric influences in multispectral optical remote sensing data |
| CN1831555A (en) * | 2005-03-10 | 2006-09-13 | 中国煤炭地质总局航测遥感局 | Processing method for local correction of remote sensing image |
| CN1710379A (en) * | 2005-07-05 | 2005-12-21 | 华东师范大学 | Atmosphere correction method of airosol optical thickness of aeronautical high-spectrum remote-sensing inversion boundary layer |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909260A (en) * | 2017-11-10 | 2018-04-13 | 浙江省地理信息中心 | A kind of natural resource assets are left one's post audit commenting method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102495405B (en) | 2013-06-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111709981A (en) | Registration method of laser point cloud and analog image with characteristic line fusion | |
| Münzinger et al. | Mapping the urban forest in detail: From LiDAR point clouds to 3D tree models | |
| CN114091613B (en) | Forest biomass estimation method based on high-score joint networking data | |
| CN107688777A (en) | A kind of urban green space extracting method for cooperateing with multi-source Remote Sensing Images | |
| CN110988909A (en) | TLS-based vegetation coverage determination method for sandy land vegetation in alpine and fragile areas | |
| Defourny et al. | Global land-cover map validation experiences: Toward the characterization of quantitative uncertainty | |
| CN108898070A (en) | A kind of high-spectrum remote-sensing extraction Mikania micrantha device and method based on unmanned aerial vehicle platform | |
| CN106485718A (en) | One kind overdoes slash recognition methodss and device | |
| CN112800827A (en) | Hyperspectral image classification experimental method | |
| Abdulkareem et al. | Accuracy assessment of digital elevation models produced from different geomatics data | |
| CN109359631A (en) | A kind of Sea Ice Types Classification in Remote Sensing Image method based on convolutional neural networks | |
| Li et al. | Modeling forest aboveground biomass by combining spectrum, textures and topographic features | |
| Nkomeje | Comparative performance of multi-source reference data to assess the accuracy of classified remotely sensed imagery: Example of Landsat 8 OLI Across Kigali City-Rwanda 2015 | |
| CN102495405B (en) | Evaluation method of TM/ETM (thematic mapper/enhanced thematic mapper) and image-based atmospheric correction product quality | |
| Tian et al. | A Process-Oriented Method for Rapid Acquisition of Canopy Height Model From RGB Point Cloud in Semiarid Region | |
| Zheng et al. | The potential of multispectral vegetation indices feature space for quantitatively estimating the photosynthetic, non-photosynthetic vegetation and bare soil fractions in Northern China | |
| Hamed et al. | Digital Change Detection and Map Analysis for Urban Expansion and Land Cover Changes in Karbala City | |
| Li et al. | Application of object-oriented classification with hierarchical multi-scale segmentation for information extraction in Nonoc Nickel mine, the Philippines | |
| CN107505265A (en) | A kind of method for performing remote-sensing monitoring on soil moisture | |
| Saba et al. | Co-seismic landslides automatic detection on regional scale with sub-pixel analysis of multi temporal high resolution optical images: Application to southwest of Port Au Prince, Haiti | |
| Apostol et al. | Data collection methods for forest inventory: a comparison between an integrated conventional equipment and terrestrial laser scanning | |
| Lu et al. | Fractal theory based stratified sampling for quality assessment of remote-sensing-derived geospatial data | |
| Huang et al. | A 10 m resolution land cover map of the Tibetan Plateau with detailed vegetation types | |
| Jeganathan et al. | Comparison of MODIS vegetation continuous field—Based forest density maps with IRS-LISS III derived maps | |
| Tang et al. | A Geometric Location Matching Method for Validation of Satellite Products: A Case Study for Albedo |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |