00:01So, this next session is going to be about imagery…

00:03…what's happening as we move forward, and hopefully many of you saw the plenary.

00:07Lots of exciting things going on.

00:09You probably got a little bit of a glimpse of some of the things that are actually happening at 10.1.

00:14Before talking specifically about 10.1, I just want to quickly talk about some of the imagery-related things…

00:20…that are coming up between now and when 10.1 gets released.

00:25These have been mentioned before. One of them is the extending off the World Imagery Services that we have.

00:31This is like this sort of background imagery of the world.

00:35As you heard, we're adding about 50 million square kilometers of GeoEye data to that.

00:40We recently released the Landsat services.

00:42These are dynamic image services which have Landsat data going back 40 years.

00:49You know, they're not just a static backdrop; they are the actual…

00:52…what's sitting on the server is the actual original Landsat scenes…

00:55…and you can dynamically interact with those and do things like feature extraction…

01:00…and look at the different band combinations.

01:02Soon we're going to be releasing the world elevation services.

01:05And these are really a collection of publicly available elevation data that is available for…

01:12…that is available and what we're doing is putting those up in services…

01:17…that can be accessed through any web or desktop application…

01:21…and those applications can render the elevation data in different ways, but also directly do analysis on it.

01:28So it's an important way of extending the use of, of imagery, going well beyond imagery to different types of elevation datasets.

01:38So, what's actually happening in 10.1 in ArcGIS? The first thing is really making things easier to use.

01:45Many of you who might be using, for example, GeoEye or Digital Globe imagery…

01:51…probably in ArcGIS Desktop of…or have within Catalog…browsed to a folder of GeoEye imagery…

01:57…and come up with a sort of a list that you probably see in the bottom right-hand side.

02:02It's a list of different shapefiles and TIFF files and NITF files and, yeah, you could use it in 10.0…

02:09…but you have to know which files to combine in different ways.

02:13In 10.1, you'll see a new item appearing essentially with the metadata.

02:18It'll actually be a folder, and if you actually open it up, you will see a list of actual products.

02:23So these are products that we can create from the raw data.

02:27So simply, you can see a list of…let's say I want to have a multispectral image…

02:32…or I want a pan-sharpened image or a panchromatic image.

02:34All you can do is just basically pick that up and drag it and drop it into ArcMap…

02:39…and it'll automatically set up all the different functions…the author rectification pan-sharpening, pan, band extraction.

02:45All that will actually be set up for you. It makes it much easier to work with these…these datasets.

02:52So, what we're also doing as part of that is extracting much more meta information from the satellite imagery.

03:00For example, the actual frequencies of the different bands, and no longer is it just a list of band 1, band 2, band 3…

03:06…but we actually know which is the red, green, and blue band, and we have meta information about the frequencies used.

03:13Also we're standardizing meta information such as sun angle and cloud…

03:18…which unfortunately has not been standardized in the industry…

03:20…so we're making it easier to combine different data sources and work with different data sources.

03:27Another important one is the image enhancement.

03:29Many of you probably have opened imagery and you get something looking like that.

03:35It's probably the wrong band combination. It looks not particularly pleasing.

03:40What we're doing is using the meta information that we know and also some techniques that we're analyzing the imagery…

03:45…and essentially automatically…maybe that screen shows better than that screen…

03:50…but automatically determining the optimized enhancement parameters…

03:54…the lookup tables that need to be applied to the imagery…

03:57…and therefore getting much clearer, crisper imagery, and without actually losing information…

04:01…for example, in the shadows and highlights.

04:03So that's optimized color correction and we've also added, for example…

04:10…interactive stretch tools so you can very easily work with the actual lookup tables and histograms…

04:17…manipulate different components off it and directly see the output. Easier to you than it was before.

04:24Also, we're working in collections of images.

04:27Those of you who are using mosaic datasets with collections of images…

04:31…you probably have already started hopefully using the color correction tools that already exist.

04:36There're a number of improvements that we've been made to that with more experience at doing this in larger projects…

04:42…and we've identified certain tweaks to that and that's making it easier to correct these very large collections.

04:48You see on the left, the typical multitude of different colors and then after running the routines…

04:54…which essentially store in memory, or rather within the mosaic dataset…

04:59…correction masks to be applied to the imagery, so the actual processing is actually done on the fly.

05:06So as you pan and zoom around, the original pixels are processed…

05:09…and as part of that process, the data is color corrected and you get a much more pleasing, pleasing actual output.

05:18Here's another sample. This is a collection of imagery.

05:21Again, it actually has been color corrected.

05:24You'll see that though that you still see the seam lines…

05:27…or rather the edges of the footprints of the individual seams and those are just straight-line edges.

05:33You can see probably a couple of them on, on the sides here.

05:37To remove…to remove that, what we've done is include an automatic seam line detection…

05:43…so these are algorithms that actually look at the overlapping imagery and determine the optimum location…

05:48…at which a seam line should actually be generated…

05:51…generate those seam lines, and they'll store those as part of the mosaic datasets.

05:55That was there before, now the actual…you…having the mosaic data…seam lines there…

06:00…but the automatic generation was actually missing in version 10…

06:04…so we've added that back so you can…so it'll automatically determine those seam lines.

06:08And this is the same mosaic dataset, but using the seam-line mosaic method and you see…you get a continuous image.

06:16Typically what you want to do, if you, for example, want to cache imagery, is it basically removes the seams.

06:24Next thing is really getting more information out of imagery.

06:28I've mentioned that we're reading much more meta information about the seams…

06:31…that we are actually reading in and we can do more with that.

06:34And one of those is a strange word, mensuration.

06:38For those in the military, it means a lot; for most of the other people, it doesn't mean much.

06:42But, simply translated, it's really the measurement of height information primarily out of imagery.

06:49So we've added these tools and as long as the imagery has sufficient metadata, that could be, for example…

06:55…a sensor model, you know, for example, a QuickBird or IKONOS image…

06:58…which has got a sensor model associated with it…

07:01…or it has information, for example, the time of day and the sun angle.

07:05Now we can use that to provide ability to measure height. And there are various tools to do that.

07:11They appear in the Image Analysis window and you can, for example…

07:15…pick between the base and the top of a building and it'll automatically compute the height of that building.

07:21It'll also actually give you estimates of the accuracy at which that…of that measurement.

07:25And there are other ways of doing it, for example, base to shadow, or the top of the building to the top of the shadow.

07:32Those are all different ways of computing height information…

07:37…which can be used, for example, to take a footprint of an image…

07:41…a building, and then add an attribute to it to create a block model, things like that.

07:46These mensuration tools are also available through the REST interface.

07:49That means that even through an iPhone application or those, you know, sort of applications or a web application…

07:57…you can actually measure heights in images now. And there are actually two versions of this.

08:03There is an unclassified version, the one which is basically part of the core product…

08:07…but then there's also, for the defense customers, there is a classified version, which is based on the MSP.

08:13For those that have access to that, they can use that to actually do…to have the certified measurements as well.

08:21This works on various types of images. Here for example, I'm working on a GeoEye imagery.

08:26Quite a lot of the satellite imagery does actually…

08:28…is actually taken a bit of an angle and you can use that alternative that you can use.

08:34For example, oblique imagery. This is imagery from pictometry and the same measurements can be made.

08:40Any imagery where you're actually seeing the size of the buildings or the shadows of the buildings…

08:45…and if the meta information is there, we can actually use these tools.

08:49Next is data sources.

08:51We already support over 60 formats, but what we're doing is carrying on to extend that…

08:57…and there are, at 10.1, there are 16 additional formats being added.

09:01These include things like the MrSID Generation 4 that's been released recently, DOQ 2, TerraSAR-X.

09:08So formats that we didn't support previously have been added.

09:12We're also improving the raster types.

09:13A raster type is really something like a collection of a format and meta information…

09:18…and how we handle that, and that can include reading the sensor model.

09:23So for example, SPOT-5 was a model that we could read before, but it wasn't accurately georeferenced.

09:28We're actually using…we have to have that sensor model built in.

09:32We have better support for things like Landsat, and again…

09:37…more for the defense customers, those that use the CSM as a transform.

09:41That can actually be used within the system as well.

09:43So that provides the ability to georeference a lot of classified imagery.

09:48Next is full-motion video integration. A very growing…number of users are starting to connect or get full-motion video.

09:58Some of it can be from, say, drones, and things like that.

10:02Some of it is just information from the side of a road or coming from a bridge, and that can now be integrated into the system.

10:09So either the live or archived feeds can be added and individualized with the georeferenced locations.

10:18Categorical data.

10:20Another thing that we've…Esri has done for many, many years…

10:23…but wasn't actually available in the mosaic datasets, has been added to the mosaic datasets.

10:29That means that if you have data which is the result of some form of a classification…

10:34…and you have associated with that some form of a database that defines what each pixel means…

10:39…now we can actually associate the two together…

10:41…and when you serve that, you will actually get not only an RGB rendering of the classification values…

10:47…but Identify and stuff will go back to the database and actually query and return actual values.

10:53So that extends the capabilities of mosaic datasets and image services into more categorical data…

10:58…which was not well supported before.

11:01Clayton already talked a bit about lidar. Just ready to reiterate some of those aspects.

11:06We see lidar as a being a very, very important dataset that we need to be able to manage…

11:12…we need to be able to disseminate, visualize, and analyze. There are really two ways of doing that, in a way.

11:18One's LAS datasets. That's really a collection of last files.

11:23Now, plus links to constraints, things like the edges of break…you know, break lines, or edges of lakes.

11:29Mainly used in project-type scenarios and you can basically create a LAS dataset…

11:35…and then access that actually within the 3D analysis tools, but you can also open it, for example, now in Spatial Analyst…

11:42…and you can actually do spatial analysis on it.

11:44It appears to the system as a virtual raster where the system rasterizes the data in the background…

11:50…and what we're actually doing is some quite advanced caching that's going on, so that if you pan and zoom around…

11:56…the rasterization is not repeated multiple times, and that improves the speed substantially.

12:01So you can directly use the LAS dataset or you can actually use the LAS files within a mosaic dataset.

12:08And that gives us the ability to catalog the potentially petabytes off LAS data that actually exists within the United States.

12:16And many organization have many terabytes of this stuff.

12:19It's all in different projections and different locations and different…yeah…many different ways.

12:25So the mosaic dataset allows us to catalog that and then when you've cataloged it…

12:30…where you've created a mosaic dataset off it, you can define additional functions.

12:34The functions do things like, oh, I want to apply a hillshade…

12:36…or I want to apply aspect or I want to do different types of filtering.

12:40You can do that in raster-type functions.

12:43It also maintains all the meta information, so if a user connects to a service and identifies a particular point…

12:49…it'll tell you the source, where it came from, things like that.

12:52So that's all really extending the mosaic dataset and the mosaic datasets can then be served as image services.

12:59When they're accessed as an image service, they can be accessed either as a virtual elevation dataset.

13:05They can be accessed as a catalog, to query. And users can actually download the data.

13:10So this can actually be used in multiple ways.

13:12Typically, an organization may have LAS data covering a small…a project.

13:17It could be a small project, like a dam or a dyke, and they may have hundreds of those within their organization.

13:22Typically what they'd do is they'd create a LAS dataset to define that project and add things like constraints to each one of those.

13:30What they can then do is to take those projects and add them to a mosaic dataset.

13:34So, they could basically create one mosaic dataset, which contains all the LAS for the whole country.

13:41They could then add to that other collections. They might have huge collections in their archives of existing LAS data.

13:47Those can be added to the mosaic dataset.

13:49They may have managed terrains, they may have datasets that they've cleaned up, and out of the billions of points…

13:55…extracted the points that are most valuable, maybe hydrographically enforced those, and created a terrain out of those.

14:02Those can be added to a mosaic dataset.

14:04And then naturally, all the other elevations data, the NED, the SRTM, the GTOPO…

14:08…the rasterized datasets that exist could be added and what you end up with is this huge virtual digital elevation model…

14:17…that can then be served to a whole range of different web, desktop applications, mobile applications.

14:23Those applications can visualize, work with the data, everything is dynamic.

14:27They can change the different rules of the data. What they can also do is analysis.

14:32So those mosaic image services can be used, for example, in Spatial Analyst.

14:36So you can create a GP tool in Spatial Analyst and do, let's say, a viewshed, or water run-off model…

14:41…or whatever you want to do, and then those tools can be served as geoprocessing tools…

14:47…and again, accessed by those applications, which no longer need to actually bring over all that elevation data.

14:53They just make a query to the system; the sever performs the processing and returns the results.

14:58Additionally, for those more advanced applications that actually want those lidar and LAS points locally…

15:03…and there are third-party applications that may need that, you can basically tell…you can through REST and SOAP…

15:11…you can actually download the LAS datasets locally and use those in those applications.

15:16So, it's really a way of managing the vast quantities of vast data.

15:19Here's a typical example. Small example, this is actually LAS data covering Oregon…I don't know...

15:27…it's about two terabytes of LAS files and you see that in a mosaic dataset…

15:32…and if you zoom in, you can change between, I want first return or last return, and you get the different renderings of the data.

15:39Next is improved tools. A lot of improvements that we've done in the general tools of how to work with imagery.

15:44One of them is georeferencing. A lot of users want, especially for things like change protection…

15:50…you need to ensure that the two images are aligned accurately.

15:54It's something that we had for a long time, but the tools hadn't been updated.

15:59In version 10.1, we've done a lot of updating the georeferencing tools, improved the user interface, so the new interface…

16:06…it's the same ribbon that you had before, but additional tools have been added to make things much simpler to use.

16:13There's a new mode, like a dual-monitor mode.

16:15Makes it easier for those with large screens to be able to see the image and the target and pick between the points.

16:21Increased accuracy. So, the old georeferencing tools did the transforms in image space, I mean ground space.

16:29In other words, they just walked what was on the ground.

16:31That was great for your traditional scanned map. It wasn't really optimum for satellite imagery.

16:37So now we can actually do those transforms in image space and that means, basically…

16:40…we transform those measurements that people have on the ground back into image space…

16:46…and then do the transforms there and that provides much higher accuracy, especially in built-up…on areas with high relief.

16:55So, there's improvement in accuracy, and additional transforms.

16:58These are things like projective, and traditionally we've had things like affine…

17:01…and if you know affine, everything was parallel, remains parallel.

17:05Well, for a lot of datasets, you don't want things to remain parallel and typically…

17:10…for example, browse imagery, you need a projective transform output added as well.

17:14Here's another couple examples.

17:16It shows some of the tools, the ability to, for example, quickly identify the location of a point approximately.

17:23In other words, if you know the approximate point of the control point…

17:25…it'll drive you to that location and you can then find that, that manhole or whatever that point is.

17:31There are things like autocomplete features, and then the residuals…

17:35…or rather if you want to actually analyze how accurate the results are…

17:38…you can…say, I want to look at my residuals and ground space or image space.

17:42So, a lot more tools to, to really ensure that the imagery is, is georeferenced quickly and accurately.

17:50Additional raster functions. So, this is becoming more and more popular.

17:55The on-the-fly processing that we talk about is really a different way of processing to traditional processing.

18:01Traditional processing is a push model.

18:04You have a raster or an image and you push it through a process, and you push it through the next process…

18:08…and you push it through the next process, to get a final output.

18:11Your on-the-fly processing really goes the opposite way around.

18:14You have defined what you want as an output, and it pulls the pixels through the functions, back out.

18:21So those are really the, the…on-the-fly processes, which we've had for on, you know, author rectification and pan sharpening…

18:28…and we've extended those so now in 10.1, you can do better things like band algebra using the same principles.

18:34And we're seeing more, more developers actually extending the system and adding their own custom functions.

18:42Function Editor is really a way to enable users to work with a mosaic dataset.

18:48Very often, a mosaic dataset may have hundreds of thousands of different images in it…

18:52…but you want to basically select maybe a hundred or a thousand of those images…

18:56…and apply a particular function to it, or edit a function.

19:01We had some tools at 10.0 and 10.1.

19:03We've realized we've had to further extend those and make it much easier to, to change parameters.

19:09I want to select a whole load of data and I want to just change the elevation model.

19:12You know, how do you do that? Well, the Raster Editor functions enable that.

19:17They also have templates, and templates enable you to make a collection of these functions to find them as a template…

19:23…and then apply that multiple times within the system.

19:26It's really extending the ability for people to create extremely large and potentially very complex mosaic datasets…

19:36…which to the end user look very, very simple. So, it's very powerful capability.

19:42Improved management.

19:43What we want to do is improve the management of, of the mosaic datasets and one other way of doing that is analyzers.

19:50So those of you who've used ArcGIS Desktop and published a map, well, an analyzer pops up and tells you…

19:57…Huh, you've got the data in the wrong projection, or you know…

20:00…you're doing certain things in…gives you some suggestions of how that can be, how that can be improved.

20:06We've added two analyzers. One for mosaic datasets.

20:11Those will go through the data in the mosaic dataset and give suggestions of things that should be changed.

20:17It may be parts that are broken or formats that aren't optimized and it'll give you suggestions of how to improve that.

20:25Similarly, for the image services, you know, if you publish an image service…

20:29…it can analyze that and also come up with some potential suggestions of what might need to be changed.

20:35Automation tools to improve automation. These are mainly as additional geoprocessing tools.

20:42We realized that there were some tools missing in 10.0, for example to change parameters in the mosaic dataset.

20:48The UI is there, but the GP tool is not there.

20:52There are some workarounds, but additional geoprocessing tools have been added to change parameters.

20:58Synchronization is a capability of the system to monitor, or rather not necessarily monitor…

21:04…but to check if anything changed within the system and then update those mosaic datasets.

21:09A typical use of that is, for example, having a directory which contains data and if you just add more data to the directory…

21:17…it should automatically be added to the mosaic dataset. It simplifies automation.

21:22Those tools exist but have been substantially enhanced to provide all sorts of different workflows and synchronization workflows.

21:31Lastly, in serving, we've done a lot of work to improve how image services are served.

21:38One is to simplify the experience. If you've been to any of the other Road Ahead sessions on Server…

21:45…you'll see the new user interface being used to enable people to serve the imagery.

21:51Now that has been simplified with a number of scalability improvements.

21:54We're always looking on how to improve the scalability of the data.

21:58Image service caching. This enables you to take an image service and actually cache it so that it becomes a map…

22:05…and that map can then be displayed as a static background.

22:09Again, for scalability reasons, that is sometimes advantageous.

22:13It was available in 10.0, but it required you to actually go and create a map and then run a set of processes…

22:21…which were very generic, and now these are optimized specifically for imagery.

22:26Server-side functions.

22:28An app, even 10.0, an application can connect to a server and define a function to be applied.

22:34I want to apply a hillshade and I want to change parameters, but we only had eight functions.

22:40Now you can have as many as you want.

22:41You can define your own functions on the server and a web application can connect.

22:46Change parameters and the server will actually do the process being applied. Very, very powerful.

22:53Image service editing. Well, what happens if you have an image service and you have an image you want to add to it?

23:00In 10.1, through a web interface, you can actually add imagery to a service…

23:04…remove imagery from a service, change imagery from a service.

23:07In other words, the management capability, if you open it up, can be performed by multiple people…

23:13…working in a web environment and modifying the services running on a server.

23:17It's very powerful, especially as we move to a cloud-type environment. And then extending the REST.

23:22The REST is really…REST interface is the way that most developers connect with the image services.

23:28Those REST interfaces have been extended.

23:30The mensuration tools, yeah? Typically you may have an application, you want to measure heights.

23:35You can do that now within web applications as well.

23:38So that's it.