The beautiful thing about rasters is that they’re useful for both visualizing and mathematical analysis. They’re simple, versatile, and the best way to represent data on a continuous, pixel-by-pixel basis.
Here are a bunch of ways you can work with this data type—and I hope the list inspires you to think of even more possibilities. You’ll see how to重组rasters to get exactly what you need, exerting specific control over cells, bands, palettes, even the way they’re geocoded. You’ll also see many ways tointegratethem, resulting in datasets that are much more rich and illustrative. If you hang around long enough, you might even learn how to use math and expression evaluation to impress your friends.
1. Convert between data formats
There are a lot of raster formats. I don’t want to embarrass myself by guessing how many (I’ve always been terrible at those Jelly Bean Jar guessing games), but I can tell you thatFME works with more than 60他们。我们使用统计数据中最受欢迎的是地理灯/ TIFF和ECW。
要得到all the benefits of data integration,通常是值得转换程序er raster format, or translating between raster and vector, or combining rasters with point clouds, databases, CAD, GIS, or any other data type. In a world wheredata is liquid, it’s suicide to keep information locked in a single format.
2. Change the dimensions
Resample the image to your desired row/column dimensions, cell size, or percentage of the original size. Resampling is usually done to generate a smaller image, but it can also be used to make a bigger (though lower quality) one.
Unless you’re looking to make pixel art, you’ll want to get essentially the same representation of a picture (for a color image) or value distribution (for a numeric raster) when downsampling. Useful interpolation methods include nearest neighbor (fast), bicubic (best quality), bilinear (a compromise), and average 4 or 16 (good for numeric rasters like DEMs).
3.更改坐标系
Like any worthy, respectable spatial data format, rasters can be georeferenced and can come in a variety of coordinate systems.
If you need to project the graphics onto a map or combine it with other data, you can reproject it to another system. For example, KML requires WGS84 (LL84), while tiles in a Web Map Tile Service are likely in Spherical Mercator.
如果您想要真正具体,您甚至可以控制您的光栅的地理编码。例如,一些栅格嵌入有地面控制点(GCP),而不是在角落上具有地理码。您可以根据需要提取或设置GCP。
4. Combine tiles into one image
Mosaic multiple rasters into a single feature. For example, maybe you need to bring tiles together into a simple overview in a lat/long projection. This would involve a mosaic, resample (#2), and reproject (#3) transformation.
It’s like a puzzle! One where the pieces don’t really fit together!
Yes. About that. Reprojecting can rotate an image slightly, and when this is followed by mosaicking, we often end up with empty black areas in between the pieces. An obsessive-compulsive nightmare. Don’t worry. You can fix it by adding a “nodata” value, so the empty areas become transparent and pick up whatever background is underneath.
5. Compress raster files
Like putting your raster in Spanx, compressing it makes everything slimmer and trimmer. 75% is a pretty good ratio to use if you want a smaller image that’s quicker to process, but still good quality.
Many formats support compression: JPEG/JPEG2000, ECW, GeoTIFF/TIFF, Oracle Spatial GeoRaster, ArcSDE Raster, Geodatabase Raster, CADRG, and WebP, to name ten.
6. Clip to boundaries
Clip an image to keep only the pieces you need, and toss away the parts outside of your defined boundary.
For example, you can merge an ECW of California with a polygon defining the coastline, and clip away the ocean (who needs that big blue wet thing anyway?). The result will be an image of just the state.
如果你剪断许多地区,像公园boundaries, you can merge the attributes from the vector features so each clipped piece retains all the properties, like the park name. As mentioned in #4, it’s handy to define the empty pieces as “nodata” so they stay transparent, making it easy to view over a background map.
7.制作光栅图形和/或元数据的目录
Consolidate graphics and their properties into a useful place, one you can share with your mom to show her all the cool stuff you do at work.
To see what I mean, check outfme.ly/parks. That’s a catalog of a bunch of clipped images and information on each piece.
目录从不支持图像的数据库中将栅格作为二进制blob读取,然后使用一些简单的html来使一切都看起来很好。
8. Split a big image into tiles
Split a raster into a series of tiles. Maybe you want images in a particular size, or maybe you want a particular number. Maybe you need smaller pieces for processing. Maybe you need to load your raster in software that doesn’t support highly compressed wavelengths. Maybe you just like your data in really tiny pieces—like data confetti. I don’t blame you. Confetti is fun.
Whatever your situation, you have a giant image and you want it in smaller pieces. Tile it.
9. Create a web map tile service
创建一系列可通过Web映射应用程序使用的图像块,例如Google地图,Bing映射或其他Web地图图块服务。
The imagery in a WMTS is stored in tile sets, so as you zoom, you see different tile sets at different resolutions. You can create a WMTS by reprojecting to spherical mercator (#3), resampling to various resolutions (#2), and splitting into tiles (#8).
10. Texture a geometric surface
我觉得没有栅格已经实现了它的命运,直到它被用来纹理表面。首先,Rasters是3D世界的2D表示。其次,3D模型通常是普通和织物的。也没有像它一样有帮助。在3D模型上覆盖光栅通常是一个启发整合。
假设你有一个轮廓的ESRI Shapefile,以及MRSID Orthophoto。您可以通过基于高程将轮廓转换为3D来创建DEM。当生成为锡并与原子光电子集成时,您最终通过具有纹理表面的更有用的3D模型,如右侧的糖果山。
Draping a point cloud with RGB and DEM rasters也导致一些伟大的表面模型转换(我们最受欢迎的一个LiDARprocessing tasks).
11. Attach an image as an attribute of a feature
A lot of formats, likeGeodatabase或Excel,允许图像附件。
This is great news for interoperability, because you can make a raster out of anything. Photos, scans, statistical data, charts, illustrations, satellite imagery … One time I demoed an enterprise raster reader/writer using a picture of a pony.
任何可以进入栅格的任何东西都可以以支持图像附件的格式。
12. Restructure the bands
Ok, now we’re getting into some more powerful stuff. (Puts on sunglasses.)
If you have multi-band data and you need to convert to a format that doesn’t support it—for example, if you have an intense 8-band format but all you want is an RGB overview—you can remove bands you don’t need.
您还可以重新介绍或添加频带,例如以具有透明度的RGB从RGB重新介绍,或删除Alpha Band,以便使用普通RGB。
13. Vectorize a raster based on values
One way to convert raster to vector is to create a polygon for each contiguous area of pixels with the same value. We call this classifying a raster.
In the vector image on the right, polygons were generated by rounding off the nearest 25 meters in the elevation model before classifying. NURBS were then made from the polygons, resulting in a vectorized contour map.
Either that, or someone put a rainbow in a blender. I’m not sure.
14. Make the bitmap realistic with hill shading
If you work with elevation data, you’re probably familiar with the plain, black and white graphics where high elevations are white and low elevations are black. Sorry, but even 3D glasses can’t make that stuff look 3D. Trust me, I tried.
Hill shading offers a real-world rendering of what the picture actually looks like. This is particularly useful for visualizing terrain when putting together a cartographic product.
15. Colorize point clouds
数据集就像香料,直到将它们混合在一起,你不经历完整的味道。
Overlay a point cloud on a georeferenced raster to color the points.
结合栅格和点云使LIDAR可视化更加令人欣喜。看看我的博客帖子14 Ways to Take Charge of LiDAR Datato see many ways to integrate point clouds. Some of the most useful, beautiful outputs involve rasters.
16. Pixel by pixel expression evaluation
Perform calculations based on the value of one or more cells. For example, you can calculate the slope or aspect (direction of slope) for each cell, or extract the extents, or detect changes between two inputs.
You could also modify pixel values based on an algorithm. Yes, friends, if these pixels store RGB values, then this is just a fancy way of saying “make your own Valencia or X-Pro II”. Free spork to anyone who uses raster expression evaluation to make their own Instagram filter.
17. Layer rasters by copying cells or using semi-transparency
组合两个栅格以查看两层的信息。您可以通过仅从其中一个图像(例如,白色和黄色道路)或通过添加透明度(#12)来完成此操作。Alpha Band允许您在覆盖它们时平滑融合图像。有点像数字铁上的转移纸。
To see how to accomplish all of the above tasks with FME, visitsafe.com/raster. I recommend you watch this October 2013webinar recordingfor great live demos, including impressive scenarios that combine several data sources (including WMS), advanced transformations, and mathematics. We also have a huge collection of articles in ourFME Community帮助您完成光栅需求。
What kind of translations and transformations do you do with raster data? What do you find the most challenging?
天安华纳
天坛is a Senior Marketing Specialist at Safe Software. Her background in computer programming and creative hobbies led her to be one of the main producers of creative content for Safe Software. Tiana spends her free time writing fantasy novels, riding her horse, and exploring nature with her rescue pup, Joey.
Ha Ha! 14 out of 17 – I am almost there!
Of course it is missing things like running interpolation processes, raster calculator, finding edges, time lapses, etc…
ruz,我认为17人中有17人让你真正成为罗斯塔的主人!谢谢你指出那些额外的功能。有趣的是,阅读所有先进的工作流程所需的数据所需的数据。亚搏在线
Morning All,
Does anyone have an example of normalizing raster imagery with FME. It’s some simple band math but I was wondering if someone had already done this in FME.
Regards.
Jeremy.
Hi Jeremy,
我假设通过归一化,你的意思是拉伸像素值以适合频带范围。您可以用几个变形金刚执行此操作:
1. You can use a RasterBandMinMaxExtractor to extract the range of values present in the band.
2. Using the Offsetter to offset the X and Y of an raster will move it, but offsetting the Z will offset the pixel values. Set the Z offset to -(minimum data value) to move the pixel values to the bottom of the band.
3. Like the Offsetter, the Scaler will scale the pixel values by the Z scale factor. Set the Z scale to (maximum possible band value / (max data value – min data value)). This will expand the data values to fill the band range.
If you would like a workspace example using your own data, please contact our Support crew through//www.baooytra.com/support, and we would be happy to help you.
Kind regards,
Dave Campanas
Product Specialist
Hello,
I’m trying to achieve nr 6 “clip to boundaries”. I’ve got a .tif-file which I’m clipping with the .shp-file of a country. That’s simple enough with the clipper transformer. What I’m struggling with is setting empty pieces to ‘nodata’ and getting transparency around the country. At the moment it’s all filled white. Which other transformer do I need to get transparency? I’d very grateful for all suggestions.
Kind regards,
Teresa
嗨特蕾莎,
It sounds like you need a RasterBandNodataSetter transformer after the Clipper, to set the background value to nodata.
Clipping and transparency is discussed in our raster webinar at about 43 minutes://www.baooytra.com/webinar/13-ways-to-avoid-a-raster-disaster/You can also download the workspaces from that webinar.
Please feel free to send us your workspace and source if you’re struggling. Our support team will be happy to help.http://fmepedia.safe.com/knowledgeSubmitCase?
干杯,
天坛
Hi Tiana,
感谢您的回复和有趣的链接。我用RasterBandNodatasetter变压器尝试了它,但从图像中删除了每一件的每一件。这可能是由于我的原始数据是带有调色板的栅格。我试图用rastedpaletteremover在剪辑之后和一个rasterpalettegenerator一起解决这个问题,但同样的问题仍然存在 - 现在是黑色的。调色板都搞砸了。
Something I thought would be quite easy to do, appears to be much more difficult….
一旦我用我的主管清算了它,我会向同事发送我的数据/工作区。也许他们可以解决问题。
Regards,
Teresa
[…] together to make photos, and those photos capture all the wonderful memories we hold most dear! Raster images can become finicky to use though if not formatted correctly for your print or web needs. In […]
Hello,
I am wondering how can I find properties like the size of an image pixel width * pixel hight?
我必须用变形金刚或用pythoncaller找到这些,
Any suggestions?
干杯。
Paal.
Got it, RasterPropertiesExtractor.
第7部分中的链接不起作用。栅格目录。这对我来说非常有用。
Hi Rob,
Thanks for letting me know. We’ll look at rehosting it on our FME Knowledge Center. For now, here is the source data in Dropbox:https://www.dropbox.com/sh/jh6c1vfpo97gwnt/AAAMW-AjKCnV2-z-zwNSCKdsa?dl=0Parks.html and the images directory are what you need for the output catalog to be shown. The FME Workspace and source data are also there.
If you want help setting this up for your own data, please don’t hesitate to contact us at//www.baooytra.com/support/report-a-problem/
干杯,
天坛
Is in the figure in 2 DEM resampled from 25 to 10 or from 10 to 25? I believe the later.
谢谢你的评论。图片显示左边的25×25光栅的表示,右侧为10×10光栅,因此在这种情况下,通过将25到10重新采样来取决于我们的意思。
干杯,
天坛
Hi,
I like to have my nodata areas filled up by interpolation… how can that be done?
最好,
Harmen
Hi Harmen,
If your rasters are numeric, we recommend using FME’s RasterDEMGenerator transformer. If you’d like help, don’t hesitate to reach out to our experts team://www.baooytra.com/live-chat/
Thanks,
天坛