I have a satellite image of the Suomi NPP satellite (nighttime lights).I have use ArcScene to create a 3D model.I exported that model in a .wrml format and I imported it into FME Desktop.I would like to know the 亚搏在线workflow for converting a .wrml file to .obj.I am totally new to FME.

As all we are knowthere is two type of raster: 1) color raster (e.g.satelliteImage) and 2) numeric raster (e.g.DEM).I am find in FME,there istutorial available for converting numeric raster (DEM) to polygon but I want toprecisely converting agriculture crops field which is almost rectangle shapefrom satellite Image to polygon.

Until here,i amsegmented NIR image by using new tools in ArcMapSegmentMean Shift.out put of that 8 bit RGB image which has mean valueof input RGB NIR image based on spatial and spectral parameter.Now inFME,i want to make pure shapefile of natural rectangle shapes like field inimage.i think there is fme raster expression which can done my job.Nowin FME i have to design expression based on segmented raster value RGB and onceagain they marge apropriyetly and make pure one rectangle segment..Lastusing RasterToPolygonCoercer we should achieve our goal.

Attached file "Sample_Data.rar" has 2 sample file: 1.Originals Image (NIR_Originals.tif) 2.Segmented Image (Segmented.tif).Thanks

sample-data.zip

As part of our role in providing humanitarian aid,we need to identify potentially thousands of shelters/tents in refugee camps with ghe aid of current satellite imagery (in the future possibly from drone imagery).It would be sufficient to identify their existence and capture a point feature - perimeter would be a bonus but not absolutely necessary at the moment.

Doing verification on automatically generated output would be much faster than the current options of manually digitising elements or field data capture.

Is there anyone who has experience in something comparable?I have virtually no experience with raster data,so thought I'd see if anyone in the FME 亚搏国际在线官网Community has any inputs/thoughts on this.

Thanks in advance.

It might be interesting if FME could read data from the Pleiades constellation of satellites.I think AIRBUS has a API to their imagery.

I wonder how others would do with this imagery once read into FME?

I wonder what aspects of the Pleiades / AIRBUS imagery are most interesting?

Start:Tutorial: Reading and Transforming Satellite Imagery in FME

Introduction

This is the last part of theReading and Transforming Satellite Imagery in FME tutorial.This exercise is fairly advanced.It will cover bit masks,bitshift and logical operations.Here we will take a look at some of the supplementary information which is collected and calculated by Landsat to help with the data quality assessment (QA).You can read more about working with Landsat's QA in thisguide from Landsat.

Downloads

landsatmask.zip

Getting Setup

Start by downloading and extracting LandsatMask.zip.Inspect LC80470252014258LGN00_BQA.TIF in the Data Inspector.Alternately,used what you've already learned in previous exercises to find your own Landsat QA raster (either with the Landsat-8 on AWS reader of theLandsat Sentinel Selector).Inspect the pixel values and note that the values are discreet,similar to what we created in the previous palette exercise,however,you should also notice that there is no palette.

The integer numbers here are really just representations of binary strings where each 0 or 1 (or pair) has a specific meaning.

Using a simple 4-bit example:

Integer - Binary String     0 - 0000       1 - 0001     2 - 0010     3- 0011     4 - 0100

We can say,for the example,that the first bit (0 or 1) from the right represents:

Data Present: 0 = nodata,1 = data,the second bit can mean snow,the third - cloud,the fourth - water.

If we look at the integer '3' above we can say the following based on the 0011 binary sting: Going from right to left - Yes we have data (1),yes,there is snow detected (1),no clouds (0) and no water (0).

Here is a handy online service which canconvert decimal to binary and back.Try using it with a few of the pixel values from the QA image.

Creating a Landsat Bitmask palette

1) Add the mask file,LC80470252014258LGN00_BQA.TIF,to workbench using the GeoTiff reader (for faster processing you can use the .ovr file)

2) Add a RasterInterpretationCoercer to convert the GREY16 band to UINT16

3) Add an UINTasBINPaletteAdder custom transformer from FME Hub

4) Right click on the UINTasBINPaletteAdder and click edit to see what it does,there is a lot here,but in particular take a look at AttributeCreator_7.Here we see a pretty scary looking expression:

@Evaluate(@Evaluate((@int64(@Value(_elevation))>>15)&1)@Evaluate((@int64(@Value(_elevation))>>14)&1)@Evaluate((@int64(@Value(_elevation))>>13)&1)@Evaluate((@int64(@Value(_elevation))>>12)&1)@Evaluate((@int64(@Value(_elevation))>>11)&1)@Evaluate((@int64(@Value(_elevation))>>10)&1)@Evaluate((@int64(@Value(_elevation))>>9)&1)@Evaluate((@int64(@Value(_elevation))>>8)&1)@Evaluate((@int64(@Value(_elevation))>>7)&1)@Evaluate((@int64(@Value(_elevation))>>6)&1)@Evaluate((@int64(@Value(_elevation))>>5)&1)@Evaluate((@int64(@Value(_elevation))>>4)&1)@Evaluate((@int64(@Value(_elevation))>>3)&1)@Evaluate((@int64(@Value(_elevation))>>2)&1)@Evaluate((@int64(@Value(_elevation))>>1)&1)@Evaluate((@int64(@Value(_elevation))&1)))

While it looks scary it is actually a concatenation of very simple expressions which usebit-wise operations.Taking one part of this as an example we can find the following:

@Value(_elevation))>>3)&1

_elevation in this case is our pixel value for this example let's say we have 216.This corresponds to a binary value of ‘11011000' (You can check this with the converter above).

The next part (216 >> 3) performs alogical bitwise right shift operationwhich shifts the binary to the right three places so 11011000 (or decimal 216) becomes 00011011 (or decimal 27).

The final part (27 & 1) compares the binary strings of each using thebit-wise Logical AND operation.We compare 11011 (27 without the leading 0's) to 00001.This results in a value of 00001 (or just 1 without the leading 0's).

Putting all of the expressions together we get either a 0 or a 1 for each of the 16 expressions,the concatenated result is the binary string of the original pixel value.The binary number is then passed into a StingFormatter witch pads the number out to 16 digits by creating leading 0's where needed.

NOTE:For a simple example you can check out suppl_SimpleBitshift&LogicalOperations.fmw.Experiment with this workspace and use the online binary converter to check the results.Try testing out some of the other bit-wise Logical operators to see how they work (- & | ^ ~).

Each bit represents the following:

You can also check out thispdf guide from Landsatfor more information for more information on how to interpret the bits.

5) Inspect the mask with palette and analyze what some of the strings mean.Connect an inspector and/or logger to the Custom transformer and run the process (the log window will show the palette).

6) Let's add some color to this.Add an AttributeCreator,and create a palette by copying the following text or typing RGB triads for some values:

RGB241 0,0,020480 0,0,020512 0,255,020516 200,200,25523552 127,127,12728672 255,127,25528704 255,127,25536864 127,127,12736896 0,255,039936 127,127,12745056 255,127,25553248 255,127,053280 255,127,061440 255,127,0

7) Add the palette with RasterPaletteAdder,run,and inspect the results,edit and experiment with a few values to show the difference.Compare the mask with the preview image to see how accurate the QC band is.Your final output should look something like this:

Processed Bit Mask:In the feature information window on the right we can see that the raster has two palettes,one containing the binary string and once with the RGB colors we created.The majority of the image is green which corresponds to a high confidence of vegetation,we also see some detection of snow or ice represented by the greys and in the top right region we can see some pink which represents the presence of Cirrus clouds.

The final workspace should look something like this.You can also open up LandsatMask.fmw from within the .zip to check the parameters.

Start:Tutorial: Reading and Transforming Satellite Imagery in FME| Next:Raster bit Masks: Making Sense of the Numbers

This article is also part of the tutorialHow Do I Do That in FME|Next: Neighborhood Functions

Introduction

This article is part of theTutorial: Reading and Transforming Satellite Imagery in FME

In this article we will be calculating theNormalized Difference Vegetation Index (NDVI).In the process,we will learn how to build conditions in the RasterExpressionEvaluator as well as how to create rasters with palettes.NDVI is based on the ability of plants to consume visible light radiation for photosynthesis (mostly in blue and red parts of the spectrum,this is why we see our plants mostly green,this color is mostly reflected).Near-infrared light,on the other hand,cannot cause the proper reaction in cells,and hence,is being reflected (otherwise it will overheat the plants).The more infrared light is reflected and the more red light is consumed,the healthier the plant is.

This article assumes you have completed the previous article in the tutorial:Making RGB images with Sentinel data.If you have not,please complete the exercise in the first part of the article to download the appropriate images.

Downloads

Completed workspace (without source data):create-palettes.fmw

NDVI Calculator

1) Create a new workspace and call it NDVICalculator.

2) Add a JPEG 2000 reader and select the preview images of the red and NIR bands (B04 and B08).Under Parameters,be sure the Feature Type Name is set to 'From File Name'

3) First add in a RasterExpressionEvaluator,set Mode to 'Two Rasters'.The Red band should go into the 'A' port and the NIR should go into the 'B' port.Use Auto for interpretation and for expression,use:

(B[0]-A[0])/(B[0]+A[0])

RasterExpressionEvaluator:This expression calculates the NDVI which is a ratio of reflected NIR to reflected Red light.

4) Run,inspect results,notice now that the band ranges of the resulting pixels falls between -1 and 1 and our interpretation has changed to REAL64.

5) Add another RasterExpressionEvaluator.In the evaluator set the interpretation to UINT8 and copy and paste the following for the expression:

if(A[0]<=0,0,if(A[0]<=0.3,1,if(A[0]<=0.6,2,3)))

RasterExpressionEvaluator_2:here we use a series of nested ‘if' functions to build a conditional expression to begin building our palette based on our calculated NDVI.This will keep only four raster values - 0,1,2 and 3

6) Add an AttributeCreator and use the text editor to create a _palette attribute as is shown below.The colors below are,navy blue,yellow,bright green and dark green but you can set the RGB triads to what you think is appropriate:

RGB240 0,0,641 255,255,02 0,255,03 0,127,0

_palette Attribute:Here we are defining the interpretation of the palette as RGB24 and are assigning each of the integers from 0 - 3 a RGB color.

7) Add RasterPaletteAdder and add _palette to the raster

8) Run and inspect results,it should look something like this:

Final Image:You will likely have a different image to what is seen here,but make sure you take note of the palette information displayed in the Feature Information window

The final workspace should look like this:

Final Workspace:Do you have something different?Download the final workspace at the top of this article to compare

Start:Tutorial: Reading and Transforming Satellite Imagery in FME| Next:Raster Calculations and Raster Palettes

Introduction

This is the third article in theTutorial: Reading and Transforming Satellite Imagery in FME.This article will run you through the process of creating an RGB image from single band raster images collected by Sentinel-2 with FME.Also demonstrated is an FME Cloud provided web service which is excellent for previewing images from both Landsat and Sentinel sources.We will also show how you can quickly and easily create a 3D surface with the satellite image draped over it by taking advantage of the AWS hosted Mapzen global Digital Elevation Model (DEM).

Downloads

Complete workspace (without source data):geotiff2obj.fmwt

Select,Download,and Inspect Sentinel Data with FME Cloud service

1) Using FME,we can build a 亚搏在线workflow that does the cloud/data filtering for us.Go to Landsat/Sentinel selector page:http://demos.fmeserver.com/landsat-sentinel-selector/.Using the tool find and download images from a location of your choice.For the purposes of this exercise select just the Sentinel-2 provider.

First select location.

Try to select an image where there is some green vegetation (otherwise the NDVI will be not visible).

Then select an image with green vegetation.

2) Download bands 2,3,4 (BGR) and 8 (NIR) AND their corresponding previews of the same bands for the same tile.In the image below,this involves downloading the regular high-resolution B02.jp2,B03.jp2,B04.jp2,B08.jp2,and from the preview folder,the previews of B02.jp2,B03.jp2,B04.jp2,B08.jp2.

Finding the files to download:The above 亚搏在线workflow should help you get to the Sentinel-2 images we want to download for this exercise.Here you can also see the Tile name from the online folder/URL path.In this case the tile name is '10UEV'.

3) Make a folder using the tile name,put all hi-res band files into it.The tile name is located at the top of the Sentinel public image browser,in the image above the 10 > U > EV results in a folder name of 10UEV.

4)使一个文件夹内的瓷砖文件夹,call it "Preview",put all previews into that folder (we will use the previews to test while building the workspace).

5) Also,while on the download page,open up the JSON metadata files and take a look at cloud and data coverage.

6) Inspect a few of the downloaded files with the FME Data Inspector using the JPEG 2000 (GeoJP2/GMLJP2) format.Please note the pixel values and interpretation (GRAY16).You will also notice that the images appear mostly black.If you look closely,however,you may notice that there are come areas which appear a little lighter.The next part of the article will shed some light on how to 'brighten' these.

NOTE:We could have also used the Sentinel-2 on AWS reader to collect these images as opposed to this manual process,however,doing it this way provides a little more insight into how the images are stored online.

Making RGB images with Sentinel data

Now that we've got the images downloaded onto your machine we can build a 亚搏在线workflow that will combine the bands together and 'brighten' the image so it looks more like what we should expect.

1) Open Workbench,add a JPEG2000 reader,read all of the downloaded preview files.Make sure the feature types are made from File Names.Previews allow very fast processing during the creation and testing stages.For final images,only the path to the source files should be adjusted.

2) Once the four feature types appear on Canvas (bands 2,3,4,8),add annotations telling what they represent (Band 2 - Blue,带3 -绿色,Band 4 - Red,Band 8 -Infrared) - this way we always see,which bands we currently use.

3) Let's experiment,what should we do with the bands?We can start by adding a RasterBandCombiner.Connect all RGB bands to it,run and inspect the result,can you tell what's wrong?

• The image is too dark,the bands are in GRAY16,so the DI will not know that it should display them as color bands.

4) Let's change the interpretations using RasterBandInterpretationCoercer for each band before combining (blue to Blue16,green to Green16 and red to Red16),run and inspect,can you tell what's is wrong this time?

• The image is still too dark.This is because the bands contain information about how much light each pixel on the sensor received,the values are not stretched to the whole available range of 16 bit

5) We can change the brightness with RasterExpressionEvaluator.Add a RasterExpressionEvaluator after the RasterBandCombiner.
Create a common multiplier published parameter,MULTIPLIER,set by default to 20:

Multiplier parameter settings.

Add the following expressions for each band:

RasterExpressionEvaluator:Here we are increasing the brightness of the pixels and changing the interpretation of each band from 16-bit to 8-bit.The MULTIPLIER determines how much brighter to make the pixels.We are dividing by 255 to get the pixel values into the 8-bit range.8-bit color bands can have pixel values up to 255 (or 2^8-1) whereas 16-bit color bands can have values up to 65535 (or 2^16-1).

6) ReRun,inspect,Better?Maybe not.

• a) If the rasters are not read in a correct order,the picture will be bright and with a good contrast,but the colors will be peculiar - for example,blue might be used instead of red and red instead of green.

7) Add an AttributeCreator after each of the RasterBandInterpretationCoercer's and create a numeric sort attribute,for example _sort with a value of 1 for Red,2 for Green and 3 for Blue.Feed them all into a Sorter before the RasterBandCombiner.Make sure Red has the lowest value,Green - second lowest,Blue - the highest.

Sorter parameters

8) Run and inspect.Now the output should look pretty good.Try adjusting the Multiplier for even better results.Real color balancing requires much more and it is not covered in this course.

9) What if we want to make an infrared image?Connect Infrared to red stream,red to green stream,green to blue stream,disable the old connections and re-run.Here we are sort of shifting all of the data higher in the visible spectrum.If you have picked an image with a lot of vegetation you should see quite a ‘red' image as trees reflect a lot of infrared light.

InfraRed satellite image

10) Add a GeoTIFF writer,set Raster File Definition to Automatic and enter the raster tile for the Raster File Name.Run the translation to make a local RGB image - we can use it for further processing or keep the current 亚搏在线workflow.

11) (Optional) Switch to the hi-res rasters (adjust the file path by removing "Preview" part) for making a hi-res image.For further experiments,be sure to switch back to previews.

Note for coordinate system information,it is necessary to use the hi-res raster images.

DEM + Satellite image + FME Magic = 3D Digital Surface

In this exercise we will take advantage of a custom transformer which pulls down and clips a subset of the Mapzen's global DEM.You can read more about the servicehere.We will then use the DEM to create a 3D surface and add the previously create color satellite image as an appearance to the surface.

1) Add MapzenAWSDEMDownloader from FME Hub to your existing workspace after the RasterExpressionEvaluator.It is important here that the image you are using has a coordinate system otherwise the workspace will fail.Please note that the hi-res images are tagged with coordinate system information while the previews are not.

2) Set some low resolution (zoom level 8) and run and inspect.

3) Increase zoom level to 10.You may need to increase the number for the 'Maximum number of tiles to download'.Save the output DEM as GeoTIFF format (or just use the transformer if your connection is good and don't care about traffic).

4) Add a TINGenerator,and connect the MapzenAWSDEM output port to the Points/Lines input port.Set surface tolerance to the DEM raster resolution(HINT: look at the zoom level in the custom transformer).

5) Add a Scaler to exaggerate the vertical component of the TIN Surface.Scale 'z' by an appropriate value (depends on location) - for BC/California values 2.5-4 work well,for flat areas - 10 or even more.

6) Add an AppearanceSetter,send the RGB raster (from the RasterExpressionEvaluator output port) to Appearance input port,TIN Surface to Geometry input port.Under Texture Coordinate Generation Parameters,set Texture Mapping Type to "From Top View"

AppearanceSetter Parameters

7) Run.Inspect results in DI

8) (Optional) Switch to hi-res to get the best quality final product in DI

9) Add OBJ writer,save the surface.

10) (Optional) Inspect results in Meshlab.Looks like Meshlab does not like textures bigger than 4096*4096 - resample before applying texture in FME if necessary - If you don't have it you can download MeshLab for freehere.

The final workspace should look like something the following:

Start:Tutorial: Reading and Transforming Satellite Imagery in FME| Next:Making RGB Images with Sentinel data

Introduction

In this exercise,the goal is to work with three available satellite image readers and provide some tips on dealing with their enormous datasets.Due to the number of satellite image services available,we will not read data from all of the services.Planet Data,Landsat-8 and Sentinel-2 will be discussed,and images read from Sentinel previews.

When reading satellite image formats,the FeatureReader is typically a better option than the traditional readers.Imagery providers have huge global datasets and reading using the traditional readers requires typing in the coordinates of a bounding box,which is not very convenient.With the FeatureReader,we can use a features geometry to initiate reading of a particular area,which is much easier.

Downloads

Complete workspace:readingsentinel-2onaws.fmw

亚搏在线Workflow Steps

1) Open a blank workspace in FME Workbench.

2) Add a Creator.

3) Add two published parameters in the Navigator - LATITUDE and LONGITUDE.Create one and select "number" in the type,then right-click duplicate to create the other.We will use the configuration to restrict the values of longitude to be between -180 and 180,and latitude - from -80 to 80 (we don't want to go to polar regions).To learn more about published parameters check outthis article

4) Create a point with VertexCreator using LATITUDE and LONGITUDE parameters ('x' is longitude)

5) Add in a FeatureReader,we will compare some of the available formats and the required parameters.

Planet Data

6) In the FeatureReader find and select the Planet Data format and open the parameters - some important things to know:

• Needs an API key - If you want,you cansign up for a free API keywhich gives you access to Planets Open California dataset
• Offers several 亚搏在线娱乐平台products - ortho scenes and ortho tiles (these will only show up if you have entered an API key)
• Has visual (RGB24),analytic (RGBNir64),UDM (unusable data mask) UINT8 assets,and thumbnails (these can be seen on the 'Feature Types to Read' dialog when using the FeatureReader)
• Allows a JSON filter.Using JSON filter is easier with FeatureReader.The filter conditions can be set through parameters,and a JSON statement created with JSONTemplater,and then - used from the attribute.

Planet Data Reader Parameters (as seen in the FeatureReader):Note that at the time of writing this article FME does not support all of the Asset/Feature Types,only those shown above are supported.

Landsat-8

7) Replace Planet Data with the Landsat-8 on AWS reader (here we will actually read the data):

• We choose to read either the high resolution images or just the overview
• We can choose the locations based on either Landsat'sWorldwide Reference System (WRS)or with a search envelope (the search envelope can be applied in the FeatureReader).The WRS is how the tiles are stored on AWS,so this mode allows retrieving all available dates for a tile.
• A date filter can be applied
• A quality filter can be applied
• Landsat has multiple bands in different spectral ranges.Each band is stored separately.Bands 2,3,and 4 represent Blue,Green,and Red respectively.Band 8 is a panchromatic image (grayscale),and can be used for pansharpening operations (increasing resolution of the lower-res color bands).Here is an exampleof the files which the Landsat-8 on AWS reader reads.

Useful transformers for making RGB images:

AttributeCreator (_sort attribute) => Sorter => RasterBandCombiner => RasterInterpretationCoercer,=> RasterExpressionEvaluator.(This process is covered overhere in this article)

Landsat-8 on AWS reader parameters (as seen in the FeatreReader)

Sentinel-2

8) Replace Landsat-8 on AWS with Sentinel-2 on AWS.This looks similar to the Landsat reader with some slight differences

• Two modes of reading
  • By resolution (not very convenient because the combinations make not much sense)
  • By bands - similar to Landsat
• Initiating geometry can be either in LL or in the Military Grid Reference System (MGRS).MGRS is how the Sentinel tiles are stored on AWS,so this mode allows retrieving all available dates for a tile.

9) When setting Sentinel-2 parameters,choose to read Feature Type Names as 'Bands',uncheck the MGRS box,and choose 'Latest' for the date filter.

10) In the Feature Reader set "Spatial Filter" to "Envelope Intersects" and Feature Types to Read choose PREVIEW (you can experiment with this to see other images for the same tile)

FeatureReader parameters for Sentinel-2 on AWS

11) Connect an inspector to the PREVIEW port and run the workspace with Prompt enabled to set the LATITUDE and LONGITUDE (try Lat = 49 and Long = -123).

12) Inspect the preview,there is a big chance that the image will have a lot of cloud cover.

13) Scroll up in the log window,and find the link to the first metadata XML,click on it,once it opens,search for cloud,show tag.

The finished workspace should look like this:

Finished Workspace

Start:Tutorial: Reading and Transforming Satellite Imagery in FME| Next:Working with Raster Readers for Satellite Imagery

Introduction

In this article,we will have a look at a few different rasters while pointing out some of the important aspects in each.We will examine the similarities and differences between the raster images.It is possible to categorize rasters using 4 different criteria:

1) georeferenced status
2) interpretation (band number/type/bitness)
3) scanned/photographed/rasterized
4) contents - imagery/dem/topographic/thematic

For another approach to viewing rasters,check out this article onViewing and Inspecting Rasters.Please begin by downloading and extracting the RasterPackage.zip below.

Downloads

RasterPackage.zip

Viewing and Inspecting Rasters

1)1_smartphonephoto.jpg

In the FME Data Inspector,open 1_smartphonephoto.jpg using the JPEG (Joint Photographic Experts Group) format.To inspect it,enable 'Select features and display information' and click on different individual pixels to discover information about them.

'Select features and display information' enabled.

You can also drag a rectangle on the image in View window and release to get information about the whole raster.

Draw a rectangle over the raster to obtain information about the whole raster.

Take note of the following:

• Ground coordinates and distances are in pixel units
• There are three bands RED8 (R),GREEN8 (G),and BLUE8 (B),their proportion on a pixel defines what color we see.
• Selecting the image as a whole,we can see that the jpeg image has a lot of metadata (format attributes)

1_smartphonephoto.jpg as seen in the FME Data Inspector:In the bottom left we see the pixel information including the Ground Location,and individual band values.This is a pixel from the leaf in the middle,note the high Band 0 Red value compared with the other bands.On the right,we see some of the metadata from the image which FME has picked up.Wondering what the numbers are in the band names (e.g.,红色的8)?This number represents the bit depth.A higher bit depth allows for more information or a greater range of values to be stored in a band.For 8 bit bands,we can only store values up to 255.The higher bit depth,however,usually means the file will typically become bigger in size.

2)2_Aerial.tif

Open aerial GeoTIFF (2_Aerial.tif) using TIFF (Tagged Image File Format) or GeoTIFF (Geo-referenced Tagged Image File Format) and inspect it.Make note of what is different from the previous raster:

• You should see that the image has a coordinate system
• Cell size and the ground units are in meters.
• Less metadata (feature attributes)

2_Aerial.tif:Similar to what we saw in the above example,however,now our image has a defined coordinate system and the image is georeferenced.If you have background maps enables you should see that it sits nicely over Vancouver's waterfront.

3)3_scannedTopomap.tif

Inspect the scanned image (3_scannedTopomap.tif),inspect how this image is different from the previous two:

• It has a single band (UNIT8) with just a few possible values
• When the whole raster is selected,each UNIT8 pixel value corresponds to a defined RGB palette that tells what color each value should be - this works just like ‘painting by numbers'.

A section of3_scannedTopomap.tif:Here we can see the palette information on the right,note that for each value in the UNIT8 band we can see a corresponding RGB color.

4) 4_elevations_cded.dem

Open and inspect the digital elevation model,CDED raster,(4_elevations_cded.dem) using Canadian Digital Elevation Data (CDED) format.

• The raster has a single band (INT32) with integer numeric values.In this case each value represents the elevation at that point.

A section of 4_elevations_cded.dem as seen the FME Data Inspector:Here we can see the INT32 band which has values corresponding to the elevation.The highest point I could find was 1497 m.

5) Landast LC80480252014249LGN00.tif

Open and inspect the Landsat image (LC80480252014249LGN00.tif) using TIFF (Tagged Image File Format) or GeoTIFF (Geo-referenced Tagged Image File Format).This one might take a little longer to load as it's very large.

• Once it has loaded we can see that this image has four bands RED16,GREEN16,BLUE16 andALPHA16.
• It also appears to be rotated.This is due to invisible pixels - it has transparent sections.These transparent sections are controlled by the alpha band.In this image the ALPHA16 band can have values of anywhere between 0 and 65535.A value of ‘0' will result in a fully invisible pixel and the maximum value (‘65535') will result in a fully opaque pix.Values in between this range will result in a partial transparency.

Landsat image as seen in the FME Data Inspector:Here we can see that it has four bands and one is called ALPHA16.An invisible pixel has been selected for the screenshot.Notice that the value is '0' for the ALPHA16 band,this corresponds to a fully transparent pixel.If we didn't have the alpha band here the border would look black (0,0,0).The reason this file is so large is because we have four 16 bit bands with a fairly high resolution.

Introduction

In this tutorial we will focus on working with satellite imagery.FME 2017+ has a number of new Readers which make it easy to quickly access and screen images from several popular satellite image services,including Landsat-8,Sentinel-2行星实验室。

A raster dataset is essentially a grid of pixels or cells with each cell containing specific information which can be used by a viewer to build a visual display.

Rasters can have geographic locations,contain location information,such as satellite and aerial images.Or,they can be just ‘regular' images such as those taken with your camera.Rasters can also be draped over 3D features to serve as textures of the 3D surfaces.

Each raster image will contain a number of ‘bands'.The number of bands a raster determines how many values are contained within each pixel.A color image typically has three bands,one for each color: Red,Green and Blue (RGB).The proportions of these band values in each pixel determine the color of the cell.

FME gives you access to inspect and transform almost all of the data contained within a raster dataset.As FME supports most common (and many obscure) formats,it can be a very powerful tool when it comes to raster processing.This support is now particularly applicable to satellite image services.With FME,it is possible to read data in one format,conduct some processing and output it as another format (or write to the same format).

Articles

1.Comparing Rasters of Different Formats and Structures

2.Working with Raster Readers for Satellite Imagery

3.Making RGB Images with Sentinel Data

4.Raster Calculations and Raster Palettes

5.Raster bit Masks: Making Sense of the Numbers