"Turn" erdas the basic processing steps of remote sensing image

Original address

Find the information when found, but also good. Suitable for beginners.

=======================================

1. Image Import
in the Import/export module of Erdas, the 1th, 2, 3, 4, 5, 7 bands of TM images are imported respectively, and the operation steps are
① Click the Import module to open the dialog box
② Select the type of TIFF
③media as file;
④ then select input, output file name path and file name
⑤ the 123457 bands respectively;
⑥ before this can select Session->preference, select input, Output home directory.
2. Image band Synthesis
A single-band image is synthesized in the ERDAS interpreter module to generate a multiband file with the following steps:
Interpreter->utilities->layerstack,
① in the dialog box that appears, select the bands that need to be synthesized in the import box, and each input band is added once with add;
②output File Select the path and name of the file to export.
③data type is set to Unsigned 8 bit;
④output option is set to union, check Ignorezero stats;
⑤ the operation.

3. Image cutting with shape file

3.1 shape file making Aoi file:
① Click the import icon in Erdas, the Import/export dialog box appears.
② Select the Imput,type bar select Shapefile,media Bar Select File, in Inputfile (*.SHP) to determine the shape file to convert, in OutputFile (*.arcinfo) determine the output path and name, Click the OK button and the Import Shapefile dialog box appears, click Importshapefile Now.
③ Note the output path and the output name are all English letters in this step
④ Building a topological polygon
⑤ Open Arctoolbox,data managementtools->topology->build in ArcGIS, double-click Build, the Build dialog box appears, and fill in the * in input. Path to the ArcInfo file, feature select Poly
⑥ Click the OK button.
⑦ Open a viewer window in Erdas, open the Arccoverage file and create a new AOI layer (New->aoi layer)
⑧view->arrange Layers Viewer Opens the Arrange layersviewer dialog box, right-click on the vector layer, select ShowProperties, open the Properties dialog, Select polygon and click the Apply button.
⑨ Open the Aoi toolbar in the View window, select the inner area, then click, Create the AOI, select the Aoi, select "Copyselection to Aoi" in the Aoi drop-down menu, click File->save-aoi Layeras, Save as Aoi file. Ok!
3.2 Using AOI files for remote sensing image cutting
in the Erdas Icon Panel toolbar, click the Dataprep icon, subset to open the Subset dialog box. In the Subset dialog box, you need to set the following parameters:
⑩ input file name
? Export file name (output files)
? Click the Aoi button to determine the clipping range
? Open the Choose Aoi dialog Box
? In the Choose Aoi dialog box, determine the source of the Aoi as file (or viewer)
? If file is selected, just further determine the Aoi file; otherwise, go directly to the next step
? Output data type is unsigned 8 Bit, output file type (outputs Layertype) is themetic
? Output cell band (select Layers) is 1:6 (indicates selection of 1-6 six bands)
? Output statistics Ignore 0 value, select Ignore Zero in Output stats check box
? Click the OK button
4. Image Preview
before you begin classifying, you need to look carefully at the synthesized image, determine the geographic area and elevation range covered by the Guardian, and identify the major landmark elements. Due to the differences in the original data of the Guardian, the same surface covering patches may display different colors in different periods of the Guardian film or in different areas of the Guardian film, so in the preview, you should also be familiar with the entire range of surface cover types and the changes between different figures.
① Open the Display window and load the cropped 6-channel image (4,3,2) or (4, 5, 3), (7,4,2);
② reduce the image to fit the window, browse the image, pay attention to the distribution of rivers, towns, vegetation, water, soil;
③ Select a specific area to enlarge to see the distribution of different types of surface coverings and variations in hue;
According to experience, the various types of surface coverings are characterized by the combination of 4,3,2 (RGB) bands:
A. Forests-forests show a variety of shades of brown, red, and brown. In high altitudes, mature coniferous forests are very dense brown or dark red; in the middle-low elevation, the color of the forest is changeable, from brown to red to dark green, some deciduous forests appear rust-yellow in winter;
B. Thickets and meadows--relative to adjacent forest patches, thickets and meadows appear bright and many red to light red. At high altitudes, large tracts of grassland may be light red or rust-red in summer, while in the winter it will be bluish green;
c. Lakes, rivers-lakes are usually black patches with clear borders, and rivers appear black or dark blue. In winter, the surface is frozen or covered with snow, showing different shades of purplish red;
d. Towns--apparently relatively bright gray or pewter patches, usually visible with regular gray lines (highways) through;
E. Farmland-the color of the changeable green, gray, lavender, light red patches, usually along the sides of the valley irregular distribution, in the Pingyuan District is large distribution. The paddy fields along the river channel tend to show a high moisture-rich bluish gray.

5. Image classification
5.1 Non-supervised classification
steps:
1th Step: Start unsupervised classification
Click the classifier icon in the Erdas Icon Panel toolbar, open the Classification dialog box, click the Unsupervisedclassification button, open unsupervised Classification dialog Box
2nd Step: Non-supervised classification
in the Unsupervised Classification dialog box, make the following settings:
① determine the input file (input Raster file) (Files to be categorized);
② determine the output file (OutputFile) (the resulting classification file), the file name is set to Ppprrr_yyyymmdd_123457_unsupervised_15.img;
③ Select Generate classification template file output Signature set, determine the template file name, named Ditto;
④ determine the cluster parameters (clustering Options), you need to determine the initial clustering method and the number of classes:
⑤ Default selection initialize from Statistics (free clustering according to the statistical value of the image);
⑥ determines the initial classification (number of classes) is 15 (divided into 15 categories);
⑦ Click the Initializing Options button to open the File Statisticsoptions dialog box, set some statistical parameters, generally using default values;
⑧ Click the Color Scheme Options button to open the Output Color schemeoptions dialog box, set the category Image Color properties, click Approximate TrueColor here, use RGB for 453-band compositing.
⑨ Other parameters take the default values.
⑩ Click the OK button (Close the Unsupervised Classification dialog box to perform the unsupervised classification).
5.2 Defining a classification template
(1) Steps:
Main->imageclassification->classification->signatureeditor, open the category template editor. Under Raster in the Viewer window, open the Tools icon and select Polygon Aoi to draw.
(2) Define template principles
① must be aware of the forest type, coverage, and overlap of images in the study area prior to classification to ensure continuity of the output classification.
② When you create a training area, there are subclasses for each category, and each subclass should select no less than 5 aoi areas, and the color type of the pixels within each AOI area is the same, and the jump cannot be large, that is, no noise.
5.3 Implementation of the supervised classification
① Select: main->imageclassification->classification->supervisedclassification to open the supervised Classifications dialog box.
② entering the original file
③ defining output Files
④ Determining the classification template file
⑤ Select output classification distance file is distance files
⑥ Defining a category distance file
⑦ Select non-parametric rules (Non-parametric rule) for feature Space
⑧ Selection Overlay rule (Overlay rule) is parametric rule
⑨ Select an unclassified rule (unclassified rule) for parametric rule
⑩ selection parameter rule is maximum likelihood (i.e. maximum likelihood method)
? Uncheck classify zeros check box
? OK to perform supervised classification.
5.4 Post-inspection correction
Open the two Viewer window to link (select the Link tool). And you can choose the AOI display function to check the correctness of sub-class selection. Repeatedly verify and modify the template.
5.5 Re-categorization
repeat the above steps to re-classify to achieve the best classification results
6. Image stitching
7. Classification re-coding
Classify the result image to be classified to re-encode, reduce the classification quantity. Determine the thematic attributes of each category, merge similar or similar categories with image recoding, and define classification names and colors.
①main->image Interpreter->gisanalysis->recode
② determine the input and output files;
③ set a new classification code (Setup Recode), open the thematic Recode table, change the "newvalue" field to the value (direct input) as needed;
④ Click OK;
⑤ Click OK
8. Filtering
①image Interpreter | GIS Analysis ... | Eliminate ... ;
② input file as ".. _ clp4.img ", Output file for" ... _ Elim25.img ";
③ "Minimum" select "25", (because the pixel resolution is 28.5m,25 pixels close to 2 hectares, so the minimum map spot is 2 hectares)
④ "Output" select "8bit"

9.erdas Registration Step (updated to Erdas2014)
Registration steps: Requirements
1. Open your correct image in the viewer module.
2. Find raster--> Geometric sub-module

3 Select the From image file, select the image you want to match, open
4, in the set geometric model that appears
Select a 2nd polyonial
5. In the dialog box below, there is a polynomial order, which is usually 2.
6, then apply, in close that dialog box
7 in the dialogue basket below, choose the first one .
8 There is a small dialog box, do not care about him, click on your first step to open the correct image interface, you can match the
9 in your accurate image inside Click a point, and then go to your need to match the image inside to find similar points with him, click, choose Enough 6 points, in the Gcptool that interface will appear error parameters, you can see whether the point is accurate. These 6 points are control points, so you should choose to spread out, as far as possible in the full extent of the map to find.
10. And then choose the other points as much as possible .
11, if you do not want to be so troublesome to find, you can click on any image of the right mouse button, and then select Geo. Link/unlink option, that way, you just need to make a point in an image, and the machine automatically gives you a look in the other,
However, the first six points will not appear error parameters, starting from the seventh point will appear error parameters, if the error is too large, you must be modified.
12, after you have finished, click on the 3rd icon in Geo correction tools OK
13, if you are in the middle to rest, you need to save the GCP Tools dialog box,
next time the match is opened directly on the line.
14. The total error after registration is that the RMS error must be at 0. 5 of cells in.
after work, your diagram is OK.

======================================================================

In addition: How to calculate the best band combination of TM using Erdas Makemodeler (in fact, it is very good to use ENVI)

1. The correlation coefficient matrix between the bands is calculated by using spatial modeler->makemodel of Erdas first. The key to this step is to select output to a file at the time of the matrix output (that is, to save the file with the extension *.mtx), to customize the dimension of the matrix according to the actual number of bands of your remote sensing data, for example, the 6*6 matrix will record the correlation coefficients between the six bands;
2. Collect information about each band. For the OIF index is mainly the standard deviation of each band (Stadev), if the other algorithms need additional information, can be recorded by layer info, respectively;
3. Calculates the OIF value according to the formula of the OIF exponent. This process can be a bit cumbersome, because if the band is more, the manual calculation is quite tedious. For this reason, I have compiled a simple program according to the calculation model of OIF index, written in VB, the calculation is very convenient and fast, if you need please send me an email. Go2happiness@163.com, if you do not reply in a timely manner, please do not be angry, because, everyone always has other things to do AH.

How to find the correlation coefficient matrix in envi

Very simply, under the statistical function can be implemented, note to the dialog box you need to check the parameters of the selected
Minimum, maximum, histogram, correlation coefficient, covariance, etc.

ENVI, select Basic Tools > Statistics> Compute Statistics
Then select the Calculate covariance Statistics check box by selecting the covariance image and/or the text Report check box to select the covariance, correlation, and eigenvectors matrix you want to output to an image, or a text record , or output to both.
Erdas can also be implemented in:

Modeler--model Maker
Create a raster object--create a function definition--create Amatrix object (Matrix Select Output form, intended to save the correlation as a matrix)

Double-click Raster, defined as the image you want to process;
Double-click function, defined as CORRELATION (<raster>), replaces the <raster>; in parentheses with the name of the image you select
Double-click the matrix to define the output file.

Envi in the density division , binary, gray Image color classification and other operations. Open the image, in the image of the tool menu dropdown colormap level Two menu has density division density slice, etc., here to operate.

How does the vegetation cover map rank in the Erdas? When you open the image, click the rasteroptions card in the Select Layer to add window, select Pseudo Color after the Display as drop-down box, open the image, click menu raster->attributes, pop-up window, Can see the color, click the color box you see, you can set the desired color. Alternatively, you can select multiple rows at the same time in the left row column.
===========
To the above information provided by the staff sincere thanks, these are the basic steps of many people need, but in the paper as if write on the same, the better the more the issue of the journal to avoid this procedure, even the processing method also write an old heap even don't understand the mathematical formula impostors, Basic processing software and means to avoid!!!!

"Turn" erdas the basic processing steps of remote sensing image