CN101963663A - Method for correcting abnormal data of 1B level remote sensing image of moderate resolution imaging spectrometer (MODIS) - Google Patents

Abstract

The invention discloses a correction method for 1B-level remote sensing image abnormal data of a medium-resolution imaging spectrometer. The method mainly solves the technical problem of how to correct the abnormal value of pixel points in the image and the abnormal data of the edge of the image. Through the imaging characteristics of MODIS data , find out the abnormal part of the image latitude data at the edge of each strip, correct the abnormal latitude data and image data through linear interpolation, and other abnormal numbers through the pixel points of the same name in the overlapping area between adjacent images, Make a quick fix. The present invention aims at the abnormal data in the MODIS 1B strip image, combines the imaging characteristics of the MODIS data and the overlap of the receiving area of the DVB-S system, and quickly corrects the abnormal data in the image. The method is simple, the execution efficiency is high, and the correction effect is high. good.

Description

A Correction Method for Abnormal Data of Level 1B Remote Sensing Image of Moderate Resolution Imaging Spectrometer

technical field

The invention belongs to the technical field of remote sensing image data preprocessing, in particular to a method for correcting abnormal data in 1B-level strip images of Moderate Resolution Imaging Spectrometer (MODIS) received through a digital satellite broadcasting system (DVB-S) , using the imaging rules of MODIS data and the receiving range of DVB-S ground receiving stations and using the same characteristics of the overlapping area data to realize the method of fast correction of abnormal data in MODIS images.

Background technique

The Moderate Resolution Imaging Spectrometer (MODIS) is one of the important sensors in the Earth Observing System (EOS) program. It has the characteristics of high timeliness, wide coverage, and rich spectral information. information, and has a wide range of applications in sudden natural and environmental disasters.

In order to improve the sharing level of remote sensing data in my country, with the support of the scientific data sharing project of the Ministry of Science and Technology, the National Satellite Meteorological Center has established a MODIS sharing platform using DVB-S, through the ground stations in Beijing, Guangzhou and Urumqi under the National Satellite Meteorological Center, and The six ground stations of Sanya, Tibet and the Oceanic Administration under the State Oceanic Administration receive remote sensing satellite data, and aggregate all the data to the National Satellite Meteorological Center to generate 1A/1B level data; finally, through the DVB-S system, digital video is used All the data will be packaged and uploaded to the communication satellite, and broadcast to domestic users. Domestic users only need simple video receiving equipment to use these satellite remote sensing data at a low price.

As an important part of my country's earth observation system MODIS sharing platform, DVB-S (Digital Video Broadcasting System) can provide Chinese researchers with quasi-real-time MODIS data, promote scientific research in related fields in my country, and greatly improve my country's remote sensing technology. The level of data sharing. However, due to signal interference and other reasons, in the data received by users through the DVB-S system, there are many abnormal data in individual images, mainly for the abnormality of image DN value and geographic positioning data.

In the past, for the correction of outliers in remote sensing images, the autocorrelation of adjacent pixels in space or time was mainly used, and a certain time or space interpolation method was used to correct the abnormal data in the images. However, for MODIS remote sensing images, a single pixel covers a large surface area, which is easily affected by cloud and other variable atmospheric elements, which reduces its spatiotemporal autocorrelation. Moreover, the values obtained by the above methods are all calculated values after interpolation operations, which will be different from the real values, and it is difficult to correct a large number of abnormal points or bands in a concentrated distribution with high precision.

Contents of the invention

The purpose of the present invention is to provide a kind of correction method for the abnormal data in the medium resolution imaging spectrometer (MODIS) 1B band image received by the digital satellite broadcasting system (DVB-S) for the deficiencies in the prior art , by using the imaging law of MODIS data and the overlapping area of the receiving range of DVB-S ground receiving stations, the rapid correction of abnormal data in MODIS images is realized.

The technical problem to be solved by the present invention is: the search of data abnormal areas in images, including the search of image pixel values and image edge data abnormal areas; the deletion of abnormal latitude data and the filling of deleted data; Adjacent to the MODIS image, and find out the overlapping area between the received images, and correct the abnormal data in the pixel points with the same name in the overlapping area, etc.

The technical solution of the present invention is to use the overlapping latitude data of adjacent scanning strips in the image to judge the overlapping area, delete the abnormal data and fill in the deleted latitude data and corresponding image data by linear interpolation, and at the same time use the corrected The latitude data and image data are used to match the overlapping area strips between adjacent images, and finally replace the abnormal data with points of the same name to realize the correction of abnormal data. The specific steps are:

Step 1: Read the image data and latitude and longitude data of the medium resolution imaging spectrometer

Use the functions in the hierarchical file format class library to read all the data obtained by the medium-resolution imaging spectrometer, and follow the steps below:

a), open the hierarchical file format file and initialize the interface;

b), converting the name of the scientific data set to be read into the index number of the data set;

c), open the corresponding scientific data set according to the index number of the data set;

d), define the start position and end position of the data set to be read;

e), read the information of the data set, and store the read data in the row and column corresponding to the two-dimensional array according to the order corresponding to the scan row and the scan column;

f), end access to the data set;

g), judging whether all data sets have been read;

h), end the interface, and close the hierarchical file format file;

Step 2: Find and correct the abnormal latitude data area at the edge of the image

a), split latitude data by scanning zone

The latitude data stored in the two-dimensional array, starting from the first column, divides the latitude data of the column in units of pixels occupied by a single scanning band in the direction of a single column, that is, in units of 10 pixels;

b), find overlapping latitude data

According to the order of the scan bands, extract the latitude data of two adjacent scan bands from the divided latitude data array in turn, compare and find out the overlapping part of the latitude data between the two scan bands, the overlapping part is the abnormal data scope;

c), delete overlapping latitude data

In this step, b) the previous scan band in the two scan bands deletes the latitude data of the overlapping part, and records the deleted latitude data and the row number corresponding to the deleted data in another array;

d), linear interpolation correction

Utilize the latitude data of the last line retained after the scan band of the overlapping part data has been deleted in this step c), and the data of the first line of the latter scan band, adopt linear interpolation to fill up the latitude data of the blank after deletion;

Step 3: Correction of abnormal image data at the edge of the image

a), delete overlapping image data

According to the result of b) search in step 2, directly find out the abnormal image data range in the adjacent scan band, and delete the overlapped image data in the corresponding previous scan band, and delete the image data corresponding to the deleted data The number of rows is recorded in another array;

b), distance weighted interpolation correction

Find out the corrected latitude data obtained in step 2 d) and the corresponding deleted overlapping part latitude data recorded in step 2 c) and the image data corresponding to the deleted overlapping part latitude data, and find out the corresponding image data after correction. Calculate the latitude distance between the point and the two nearest neighbors of the two deleted latitude data, and use distance-weighted interpolation to fill the deleted image data;

Step 4: Find overlapping regions between adjacent images

a) Basic information extraction of medium resolution imaging spectrometer images

Using the naming rules of medium-resolution imaging spectrometer images received through the digital satellite broadcasting system, the basic information of MODIS images is extracted from the file names of the images, including the name of the satellite on board, the spatial resolution of the images, and the imaging time of the images;

b), judging whether there are adjacent images

The satellite where the medium-resolution imaging spectrometer is located passes through the next imaging time of the same latitude area for about 1 hour and 30 minutes. Based on this, the images whose imaging time interval is within 30 minutes can be defined as adjacent images in time, and the images are analyzed. Grouped, arranged in the order of imaging time;

c), search for overlapping areas

Match the latitude data of two temporally adjacent medium-resolution imaging spectrometer images with all scan lines in the other image in accordance with the imaging sequence of the scan bands, starting from the first line of the image with an earlier imaging time , so as to find out all scan lines with exactly the same image data, and determine the range of overlapping regions between temporally adjacent images;

d), end search

Step 5: Judgment of overlapping areas

Utilize the overlapping area search result in step 4 c) to judge whether there is an overlapping area;

Step 6: Correction of abnormal image data

Start from one end of the overlapping area of the images of the two adjacent medium-resolution imaging spectrometers respectively, and compare the pixels at the same position in the same scan band sequence. If the value is normal, replace the abnormal pixel value and complete the correction of the abnormal data until all the scanning bands in all overlapping areas are compared;

Step 7: Output the result.

The search and correction of the abnormal latitude data area on the edge of the image is to judge and correct the abnormal data only by using the abnormal latitude data; since the MODIS sensor is carried on the two satellites AUQA and TERRA, and the orbits of the two satellites are different, the AUQA satellite The latitude data of the normal latitude data in the column direction are monotonously increasing; the latitude data of TERRA satellites are monotonically decreasing in the column direction; the correction of abnormal data is directly in the single column direction On the above, find the overlapping area of adjacent scanning bands, correct the abnormal data by deleting the overlapping area and linear interpolation, and correct the abnormal image data by distance weighted interpolation.

The overlapping region search between the adjacent images is based on the MODIS file naming rules received by the Digital Satellite Broadcasting System (DVB-S), extracting the satellites carried, the spatial resolution of the images and the imaging time information of the images; and in 30 minutes As a criterion for judging image groups that are adjacent in time and have overlapping areas; according to the order of scanning bands, the overlapping areas between images are searched based on the double matching method of latitude data and image data.

In the hierarchical file format (Hierarchical Data Format), that is, HDF, the unit of data storage is a scientific data set, and all data are divided according to a certain logical order, such as longitude and latitude data sets, etc., which are stored in different scientific data sets. data set.

By analyzing the distribution of all receiving sites in the DVB-S system, the present invention finds that there is a large overlap between the receiving ranges of each site, so it can replace the abnormal ones with normal pixel data by searching for the pixels with the same name in the overlapping area between images Data, so as to realize the rapid and accurate correction of a large amount of abnormal data.

Description of drawings

Fig. 1 is a flowchart of the present invention

Fig. 2 is the position of DVB-S MODIS data receiving station of the present invention and the schematic diagram of receiving range

Fig. 3 is a schematic diagram of the scope of overlapping regions according to an embodiment of the present invention

Figure 4 is a schematic diagram of the embodiment of the present invention before abnormal data correction

Figure 5 is a schematic diagram of the corrected abnormal data according to the embodiment of the present invention

Detailed ways

Referring to Fig. 1, the present invention is a method for correcting abnormal data in remote sensing images. The method utilizes the overlap of received data ranges between stations to find overlapping areas between adjacent images, and matches the pixel points with the same name, using the Normal cell values replace abnormal cell values to achieve accurate correction of abnormal data. The specific correction steps are:

Step 1: Reading MODDIS image data 10 and reading latitude and longitude data 20

MODIS image data 10 mainly uses functions in the hierarchical file format class library to read all MODIS data. The steps and functions used are as follows:

a), open the hierarchical file format file and initialize the SD (SDstart) interface 21;

b), the name of the scientific data set to be read (such as the name of the latitude data set is Latitude) is converted to the index number (SDnametoindex) 22 of the data set;

c), open the corresponding scientific data set (SDselect) 23 according to the index number of the data set;

d), set the start and end position 24 of the data set read, if the read data set is a multidimensional data set (more than one dimension), then define the start position and the end position (start, end) of the read data set edge), otherwise no need to specify;

e), read the information (SDreaddata) 25 of the data set, if the latitude and longitude data are read, the read data is stored in the corresponding row and column of the two-dimensional array according to the corresponding positions of the scanning row and the scanning column;

f), end access to the data set (SDendaccess) 26;

g), judging whether all specified data sets 27 have been read;

h), end the SD interface, close the hierarchical file format file (SDend) 28;

Step 2: Find and correct the abnormal latitude data area at the edge of the image30

a), divide the latitude data 31 according to the scanning zone

The latitude data stored in the two-dimensional array, starting from the first column, divides the latitude data of the column in units of pixels occupied by a single scanning band in the direction of a single column, that is, in units of 10 pixels;

b), find overlapping latitude data 32

According to the order of the scan bands, extract the latitude data of two adjacent scan bands from the divided latitude data array in turn, compare and find out the overlapping part of the latitude data between the two scan bands, the overlapping part is the abnormal data scope;

c), delete overlapping latitude data 33

In this step b) the previous scan band in the two scan bands deletes the latitude data of the overlapped part, and records the deleted latitude data and the row number corresponding to the deleted data in another array;

d), linear interpolation correction 34

Utilize in this step c), the latitude data of the last line retained after the scanning band of deleted overlapping part data, and the data of the first line of the latter scanning band, utilize the method for linear interpolation to fill up the latitude data of the blank after deletion;

Step 3: Correction of abnormal image data at the edge of the image 40

a), delete overlapping image data 41

In the MODIS raster image data, there is a one-to-one correspondence between the image data of a single grid point and the latitude data, so the image data corresponding to the abnormal latitude data is also abnormal. This correspondence can be used, according to b) in step 2 The result of the search directly finds out the abnormal image data range in the adjacent scanning zone, and deletes the overlapping image data in the corresponding previous scanning zone, and records the deleted image data and the row number corresponding to the deleted data in another in an array;

b), distance weighted interpolation correction 42

Find out the corrected latitude data obtained in step 2 d) and the corresponding deleted overlapping part latitude data recorded in step 2 c) and the image data corresponding to the deleted overlapping part latitude data, and find out the corresponding image data after correction. Calculate the latitude distance between the point and the two nearest neighbors of the two deleted latitude data, and use distance-weighted interpolation to fill the deleted image data;

Step 4: Search for overlapping regions between adjacent images 50

a), MODIS image basic information extraction 51

Using the naming rules of the received DVB-S MODIS image file name, extract the basic information of the MODIS image from the file name of the image, including the name of the satellite carried, the spatial resolution of the image and the imaging time of the image;

b), judging whether there is an adjacent image 52

The next imaging time of the MODIS satellite passing through the same latitude area is about 1 hour and 30 minutes. Based on this, the images whose image imaging time interval is within 30 minutes can be defined as temporally adjacent images, and the images are grouped according to the imaging Arranged in chronological order;

c), search for overlapping areas 53

Since the latitude data and image data of the image in the overlapping area are exactly the same, the MODIS data processed by the abnormal latitude data of the image edge and the image data can be used to convert the temporally adjacent MODIS image from the earlier image Starting from the first line, according to the imaging sequence of the scanning zone, the latitude data is matched with all the scanning lines in another scene image adjacent in time, so as to find out all the scanning lines with the same image data, and determine the time The extent of the overlapping area between adjacent images;

d), end search 54

Step 5: Judgment of overlapping areas 60

Utilize the overlapping area search result in step 4 c) to judge whether there is an overlapping area;

Step 6: Correction of abnormal image data 70

Start from one end of the overlapping area of two adjacent MODIS images respectively, and compare the pixels at the same position in the same scanning order. If the value of any one of the pixels is abnormal and the corresponding value of the other pixel is normal, then Replace the abnormal pixel values and complete the correction of abnormal data until all scanning bands in all overlapping areas are compared;

Step 7: Output the result as 80.

Example

Refer to Figure 1, Figure 3, Figure 4 and Figure 5, and correct the abnormal data of the MODIS image with imaging time around 6:00 on June 4, 2008. The selected MODIS data are Guangzhou, Beijing and Urumqi in the DVB-S system The MODIS data received by the ground receiving station from the AQUA satellite is an image with a spatial resolution of 1KM. According to the flow chart of the present invention, the abnormal data in the three images are corrected, and the image data and corresponding latitude and longitude data of these sites are read through step 1, and then the correction method given in step 2 and step 3 can correct the abnormal data edge The latitude data is corrected according to the linear interpolation method, and the corresponding abnormal image data is corrected according to the distance weighted interpolation method, and then the overlapping area between adjacent images is found, and the basic information is extracted from the file name of the MODIS file. The image information includes the satellites carried, the spatial resolution of the image and the imaging time of the image. In the embodiment, the MODIS image file received by the Guangzhou station is named AQUA_2008_06_04_05_48_GZ.MOD021KM.hdf. Here, the file name of the file is used as an example to illustrate from The method of extracting the basic information of the image data from the file name is to extract the name of the satellite carried by the MODIS sensor as AQUA from the character "AQUA" in the file, and to obtain the spatial resolution of the image as 1 km from the character "1KM" in the file name. The character "2008_06_04_05_48" in the file name was extracted and the imaging time of this image started at 5:48 on June 4, 2008. Calculate the imaging time intervals of different images from the extracted image imaging time, divide the intervals within 30 minutes into a group, correct the abnormal data, and finally find the overlapping area between adjacent images (as shown in Figure 4 ), at the same time, starting from one end of the overlapping area of the two MODIS images, compare the pixels at the same position in the scan zone, and replace the abnormal pixel values with normal pixel values until all the pixels in the overlapping area are compared . The image comparison before and after abnormal data correction is shown in Figure 4 and Figure 5.

Table-1: Main Technical Parameters of MODIS Moderate Resolution Imaging Spectrometer

Table-1 Main parameters of the MODIS instrument

According to data retrieval, most of the existing correction methods for abnormal data use spatial interpolation or time interpolation methods, and the corrected data is not the real value received by the sensor. method can not be resolved. For this reason, the present invention adopts the characteristics that there are overlapped areas between the receiving ranges of the receiving stations of the DVB-S system, and the values of the overlapped parts are completely the same, to correct the MODIS data. Moreover, it is possible to correct various abnormal data (image data, longitude and latitude data) at the same time.

Aiming at the imaging time of June 4, 2008 at around 6 o'clock, the data from the three stations in Guangzhou, Beijing and Urumqi were processed for image outliers. It can be seen from the effect diagram of the image processed by the method of the present invention that the present invention can well correct the abnormal data existing in other images, the correction effect is good, the method is simple, and the execution efficiency is high.

Claims (3)

1. A correction method for 1B level remote sensing image abnormal data of a medium-resolution imaging spectrometer, characterized in that the method comprises the following specific steps: Step 1: Read the image data and latitude and longitude data of the medium resolution imaging spectrometer Use the functions in the hierarchical file format class library to read all the data obtained by the medium-resolution imaging spectrometer, and follow the steps below: a), open the hierarchical file format file and initialize the interface; b), converting the name of the scientific data set to be read into the index number of the data set; c), open the corresponding scientific data set according to the index number of the data set; d), define the start position and end position of the data set to be read; e), read the information of the data set, and store the read data in the row and column corresponding to the two-dimensional array according to the order corresponding to the scan row and the scan column; f), end access to the data set; g), judging whether all data sets have been read; h), end the interface, and close the hierarchical file format file; Step 2: Find and correct the abnormal latitude data area at the edge of the image a), split latitude data by scanning zone The latitude data stored in the two-dimensional array, starting from the first column, divides the latitude data of the column in units of pixels occupied by a single scanning band in the direction of a single column, that is, in units of 10 pixels; b), find overlapping latitude data According to the order of the scan bands, extract the latitude data of two adjacent scan bands from the divided latitude data array in turn, compare and find out the overlapping part of the latitude data between the two scan bands, the overlapping part is the abnormal data scope; c), delete overlapping latitude data In this step, b) the previous scan band in the two scan bands deletes the latitude data of the overlapping part, and records the deleted latitude data and the row number corresponding to the deleted data in another array; d), linear interpolation correction Utilize the latitude data of the last line retained after the scan band of the overlapping part data has been deleted in this step c), and the data of the first line of the latter scan band, adopt linear interpolation to fill up the latitude data of the blank after deletion; Step 3: Correction of abnormal image data at the edge of the image a), delete overlapping image data According to the result of b) search in step 2, directly find out the abnormal image data range in the adjacent scan band, and delete the overlapped image data in the corresponding previous scan band, and delete the image data corresponding to the deleted data The number of rows is recorded in another array; b), distance weighted interpolation correction Find out the corrected latitude data obtained in step 2 d) and the corresponding deleted overlapping part latitude data recorded in step 2 c) and the image data corresponding to the deleted overlapping part latitude data, and find out the corresponding image data after correction. Calculate the latitude distance between the point and the two nearest neighbors of the two deleted latitude data, and use distance-weighted interpolation to fill the deleted image data; Step 4: Find overlapping regions between adjacent images a) Basic information extraction of medium resolution imaging spectrometer images Using the naming rules of medium-resolution imaging spectrometer images received through the digital satellite broadcasting system, the basic information of MODIS images is extracted from the file names of the images, including the name of the satellite on board, the spatial resolution of the images, and the imaging time of the images; b), judging whether there are adjacent images The satellite where the medium-resolution imaging spectrometer is located passes through the next imaging time of the same latitude area for about 1 hour and 30 minutes. Based on this, the images whose imaging time interval is within 30 minutes can be defined as adjacent images in time, and the images are analyzed. Grouped, arranged in the order of imaging time; c), search for overlapping areas Match the latitude data of two temporally adjacent medium-resolution imaging spectrometer images with all scan lines in the other image in accordance with the imaging sequence of the scan bands, starting from the first line of the image with an earlier imaging time , so as to find out all scan lines with exactly the same image data, and determine the range of overlapping regions between temporally adjacent images; d), end search Step 5: Judgment of overlapping areas Utilize the overlapping area search result in step 4 c) to judge whether there is an overlapping area; Step 6: Correction of abnormal image data Start from one end of the overlapping area of the images of the two adjacent medium-resolution imaging spectrometers respectively, and compare the pixels at the same position in the same scan band sequence. If the value is normal, replace the abnormal pixel value and complete the correction of the abnormal data until all the scanning bands in all overlapping areas are compared; Step 7: Output the result. 2. The correction method according to claim 1, characterized in that the search and correction of the abnormal latitude data area at the edge of the image is to judge and correct the abnormal data only by using the abnormal latitude data; the correction of the abnormal data is directly in the single column direction On the above, find the overlapping area of adjacent scanning bands, correct the abnormal data by deleting the overlapping area and linear interpolation, and correct the abnormal image data by distance weighted interpolation. 3. The correction method according to claim 1, characterized in that the search for overlapping regions between adjacent images is based on the medium-resolution imaging spectrometer image file naming rules received by the digital satellite broadcasting system, extracting the loaded image from the file name. The satellite, the spatial resolution of the image, and the imaging time information of the image; 30 minutes is used as the criterion for judging the division of image groups that are adjacent in time and have overlapping areas; according to the order of the scanning bands, the search is based on the double matching method of latitude data and image data The overlapping area between images.

Images