Previous: Mathematical Operations on Two Rasters
Overview
The highly anticipated RasterConvolver is finally here! New to FME 2018.1, the RasterConvolver is a transformer that applies a convolution filter to raster data. Convolution filters are applied using a kernel which is specified by a matrix of weights for the neighboring values. Filtering can be used to smooth, sharpen, enhance edges and other raster image manipulation operations.
Downloads
Exercise
In this scenario, you are interested in applying a convolution filter in order to sharpen a GeoTIFF image. For more information on Raster processing using the RasterConvolver, check out the Edge Detection with the RasterConvolver article or the Raster Slope Calculations as a RasterConvolver Example article.
Instructions
1. Start FME Workbench
- If it isn’t open already, launch FME Workbench.
2. Create a New Workspace
- In the Create Workspace part of the Start page, select the option to Create a New Workspace.
3. Add a GeoTIFF Reader
- Start typing “GeoTIFF” without anything selected on the canvas, then select the GeoTIFF format from the list of Readers by double-clicking or by using the arrow keys and the Enter key to add it to the canvas.
- Click on the Dataset ellipsis and navigate to the folder containing BCVANC15_O7.tif. Select the BCVANC15_O7.tif file and click Open then OK to add the GeoTIFF reader to the canvas.
4. Add a RasterConvolver
The RasterConvolver accepts input features containing a raster geometry and outputs the features after applying a convolution filter to each band.
- Add a RasterConvolver to the canvas by typing “RasterConvolver” to bring up the list of FME Transformers in the Quick Add Search. Select the RasterConvolver from the list of Transformers by double-clicking or by using the arrow keys and the Enter key to add it to the canvas.
- Once the transformer has been added, connect the GeoTIFF Reader to the ConvolutionFilter.
5. Set the Convolution Filter Type
- Once the RasterConvolver has been added, double-click the RasterConvolver or click on the gear icon to open the transformer parameters dialog.
- In the Parameters dialog, change the Convolution Filter to Sharpen 3.
- Expand the Advanced section and set the Preserve Band Interpretation to Yes.
Note: By default, the preserve band interpretation is set to No - leaving this as No would convert the Red, Green, and Blue bands of the GeoTIFF to Real64 (i.e. convert to a grayscale image).
6. Add an Inspector
- Add an Inspector to the canvas and connect it to the RasterConvolver:Output port. Once you have made the connection, open the Inspector parameters and name it Sharpened.
- Similarly, attach an inspector to the GeoTIFF reader and name it Original. This will automatically open both the original and sharpened raster dataset in the FME Data Inspector after the translation has run. Inspecting both datasets will allow you to toggle between the original image and the sharpened image to see how the RasterConvolver’s Sharpen 3 convolution filter processed the image.
7. Add an ECW Writer
- Similar to adding a reader, type “ECW” then select the ECW format from the list of Writers.
- Specify a folder for the ECW file by clicking on the Dataset ellipsis. After specifying a folder location, set the Raster File Definition to Automatic then click OK.
- Next, set the Raster File Name in the Feature Type Parameters dialog then click OK to add the ECW writer to the canvas. Here, you can also specify the Raster Compression by percentage and have the option to generate a tab file. To learn more about TAB files see the Raster Documentation.
- Lastly, connect the ECW Writer to the RasterConvolver:Output port.
8. Run the Workspace
- Run the workspace by clicking the run button on the toolbar, or by using Run > Run Translation on the menu bar.
- After the translation has run, toggle between the Sharpened and Original image to see what the Sharpen 3 convolution filter did. Try picking another filter from the RasterConvolver’s predefined convolution filters or enter your own values by setting the Convolution Filter to User Defined and entering Kernel Weights.
Results
Input
Output
Transformers
- Inspector - Sends features to the FME Data Inspector for display.
- RasterConvolver - Accepts input features containing a raster geometry and outputs the features after applying a convolution filter to all the bands.
Data Attribution
Data used in this tutorial originates from open data made available by the City of Vancouver, British Columbia. It contains information licensed under the Open Government License - Vancouver.
