StreetView Cars Collecting 3D Data on Buildings

San Francisco in Google Earth 4.3

CNET has a story "revealing" that Google has been collecting 3D data in addition to the imagery shown in the StreetView map views. This may come as a surprise to some out there, but there was a May 2007 story from the San Jose Mercury News that Google would be using technology from Stanford University to collect 3D data from the car systems This does pose the question whether the new 3D buildings added with Google Earth 4.3 for several cities used the new system. But, since Baltimore for example got the new 3D buildings, and there aren't StreetView photos for that city, I'm thinking perhaps they have yet another method of generating 3D buildings. Also, I think it would be hard to generate 3D buildings of skyscrapers using this technology. My guess is it would be better for smaller buildings - and maybe trees? - than for cities. My theory: Google will use a combination of technologies for 3D model generation - and they'll be trade secrets.

Times Square at Night in Google Earth/Maps StreetView

Thanks to Google Earth's StreetView layer, I noticed some of the new StreetView images released today have some night-time views of Times Square. See it here in Google Maps:

View Larger Map

In Google Earth (v4.3 only), if you turn on the StreetView layer and zoom down to near street level, you'll see miniature spheres showing the images in place of the icons. That's how I quickly noticed the different set of images. See more about StreetView in the video demonstration of Google Earth 4.3 here (the last half of the video shows it).

The Times Square area is a particularly good area to check out the new face blurring technique Google is using to help reduce privacy concerns about StreetView imagery. They have face detection software to help with the blurring process.

New Imagery for Google Earth - May 2008

[UPDATED 14-May 1510 ET - official details on the imagery updated added below.]

Google has released new imagery today for Google Earth. As usual, they are not revealing the locations yet - instead we get some hints of locations in the form of some riddles. I've been busy with attending the Where 2.0 conference, and haven't had time yet to check out the imagery myself.

Here's the answers to some of the riddles (thanks to help from some comments below): 1) Los Angeles, 2) Coronado Island near San Diego, 3) Houston, Tx, 4) Lisbon, Portugal, 5) Madrid, Spain; 6) Milan, Italy, 7) Melbourne, Australia, 8) Miami, Florida, 9) Isle of Man

Thanks to GEB readers for find a lot of the new imagery, but here are the official details from Google's LatLong blog:

New high resolution:

Significant amount of new satellite imagery in Ecuador, Peru, Senegal, Nigeria, Kenya, Uganda, Tanzania, India, Iran, Bangladesh, Thailand, Vietnam, Malaysia, and The Philippines. Additionally, new 2.5m imagery for part of Western Australia

Updated Imagery:
Americas:
- USA: LA, San Diego, Houston, Miami, Chicago and Milwaukee area suburbs, New York City area suburbs, much of coastal New Jersey, and Harney County (Oregon).

Europe, Middle East & Africa:

- England: Isle of Man, Suffolk

- Spain: Madrid
- Portugal: Lisbon, Guimaraes, Porto, Sevilla, Coimbra, Costa del Sol, Costa Blanca
- Italy: Milan
- France: Toulon, Montbeliard
- The Netherlands: Assen

Asia & Oceania:

- Armenia: Yerevan
- Australia: Melbourne, Darwin

Updated Terrain:
- Westport, Ireland
- Hawaii
- Puerto Rico

Google Geo Search API, ESRI and Google Collaborate

[UPDATED 1730PT: Added the video of the talk below.]

At Where 2.0 today John Hanke, the Director of the Google Geo team, had a couple of announcements. He started out by giving a report on the state of the GeoWeb as he sees it. According to Google's statistics geospatial data grew by 300% since last year's Where 2.0. He also described the progress in their geosearching which started with Google Earth and has progressed into Google Maps with more features. He said he had one announcement related to that: Google is now releasing a GeoSearch API. An AJAX API that allows you to tap into Google's geodata search indices. It will be interesting to see if you can get more results than the standard Maps/Earth interfaces and then output the results in KML. I'll link to details on their web site after they formally make this announcement.

The other interesting thing that happened during John's talk was his strategy to get more GIS data into the Where 2.0 type applications world. John went and spoke to Jack Dangermond, CEO of ESRI, to see if he could convince him to work on ways to better enable access to GIS data. He invited Jack up to the stage and Jack quite enthusiastically described his shared vision with Google to get GIS data more accessible. Apparently ESRI will be intorducing new features in the applications to facilitate KML output of GIS data. Jack said the new features will start rolling out in about 4 weeks. The implication was that it would be possible to create mashups between GIS databases and neogeography databases and tools. They discussed a bit the issues that might crop up between crowdsourced data vs. professional GIS data, but that in the long run it could be for the best. Emergency situations like the California fires was given as an example where both data types proved useful.

Here's a video of the presentation:

Where 2.0 - Day 1

Left my house near Raleigh, NC at 4 AM to fly out to California. I was a bit weary after nearly 11 hours of travel when I arrived at the Where 2.0 registration desk. But, I did manage to show up just in time for lunch! Brady Forrest of O'Reilley, Program Chair for Where 2.0, joined me for lunch and brought me up to date on events thus far. He said the morning tutorials were well attended and seemed to be quite popular (see schedule for Monday). Also, nearly half the people who registered for the conference were in attendance for the first day events. Also, overall attendance looked to be higher than last year. Several announcements are lined up for the week as well from both big and new players in the geospatial community. Looks like lots of good stuff for this year's event!

The afternoon tutorials included another four tutorials. Naturally, I chose to attend the one by Google entitled: "Searching the Geoweb: Exposing Your Geo Data to Search Engines" by Lior Ron and Mano Marks. Their session focused on how to get your geospatial data to be recognized by search engines (or at least Google). Google has already been scanning web sites for links to GeoRSS, KML, or KMZ files as well as the Google Earth Community and other known locations. They gave tips on how to improve the chances things will be indexed. A significant point is that you shouldn't make a single large file with tens of thousands of placemarks. Better to break down the content into different files. The one piece of real news in this session is that Google is now supporting a new sitemap file format for geospatial content. This means you can very specifically tell the search engines to look for a KML, KMZ, or GeoRSS file which will then be almost immediately scanned into the geosearch index. The last half of the session was dedicated to discussing how to create good KML content using various tools as well as directly coding KML. Google said there will be more of these tutorials during the Google I/O developer conference later this month.

Keep reading for some details on Ignite/Launchpad.

Links: New StreetView Stuff, WorldWide Telescope, Yahoo WOE, Dishpointer, GPSAnimator

• New StreeView Stuff - Google pre-announced new StreetView data for New York City. The data reportedly has higher resolution data. However, so far I'm not seeing the new StreetView imagery. But, those of you concerned about seeing your face appear there - don't worry - Google has used face-detection software that "blurs" the faces in the imagery. The data is also reportedly more recent. I did notice a new "More" drop-down option that lets you turn on Panoramio or Wikipedia placemarks though.
• WorldWide Telescope - Microsoft has surprised many of us by releasing a beta of their widely anticipated WorldWide Telescope application. Stefan Geens at OgleEarth calls it "stunning", and further says: "this is easily the most impressive thing I've seen Microsoft do in a long time...". I haven't had a chance to try installing it, but hopefully will before the week is out. I'll draw comparisons with Google Earth's Sky mode. Mickey says it has some bugs, but is definitely worth checking out. The New York Times has an interesting article describing both WWT and Google Earth Sky and compares them from a philosophical approach.
• Dishpointer.com - Stefan at OgleEarth also has a great review of an interesting new web site called Dishpointer. This site will calculate the angle for TV satellite antennas. More importantly, it will let you see it in Google Earth so you can turn on 3D buildings and see if you might be obscured by buildings.
• Yahoo WOE - Rev Dan Catt announced a new Yahoo capability to offer "Where on Earth IDs". Here's the Yahoo announcement. Dan focuses on the Flickr aspects of using these new ids. Ed Parsons from Google also comments positively on the Yahoo effort.
• GPSAnimator.com - Got an E-mail from some folks who have developed an application called StarTRAX which lets you visualize your GPS tracks in Google Earth. I haven't had a chance to try it out, but it sounds a lot like GPSVisualizer.

Where 2.0 2008 - Day 2 - Tuesday

Yesterday was the first full day of sessions at Where 2.0 2008. Of course, I already blogged the Google keynote by Director of Google Geo John Hanke (I've added a video of his presentation to the post). Also, note there are other videos of some of the presentations at Where available at where.blip.tv. I'm not going to have time to write summaries of all the talks from yesterday right now. So, here are a few notable summaries which I found most interesting from a Google Earth perspective.

• Microsoft Virtual Earth - The other important keynote of the day, from a GE perspective, was from the Microsoft Virtual Earth team (for a good summary of the entire talk, see Chris Spagnuolo's post). Microsoft's Vincent Tao, formerly the founder of GeoTango, gave an enthusiastic presentation - and revealed not only current progress, but some interesting glimpses at future technology they are working on. I found it interesting that Vincent claimed "VE is the largest mapping project ever in the industry". Given the amount of money have spent, and the huge volume of data they have generated (3D city models, Bird's Eye Views, as well as high resolution ortho photos and street maps) - they are certainly generating a lot of bytes of imagery. But, don't forget Google has StreetView imagery, ortho imagery, street maps, and 3D city models as well. And, Google has many layers of information not found in Virtual Earth (borders, places of interest, Gigapan photos, Panaramio, etc., etc.). Google started the process of "crowd sourced" mapping data which has generated vast amounts of data as well. So, I think it is presumptuous to say Microsoft's project is bigger. Vincent also demonstrated the impressive Virtual Earth upgrade from last month with a particular emphasis on the new "version 2" 3D cities. Currently there are only four new cities, but I heard from other Microsoft people they plan to add several new v2 cities a month soon. As I've said before, the new 3D trees they've added look awesome, and the upgraded model textures definitely enhance the look of their cities. Probably the most interesting new demonstration by Vincent was a new upcoming capability they are working on: automatically placed photo overlays. The demo appeared to offer the same 3D placement of photos as Google Earth's new PhotoOverlay (first introduced last fall). But, the big difference is that Vincent claimed they could automatically place the photo. I suspect this is using the same underlying photogrammetry technology used to generate their 3D models. This could be a huge development since putting photos in Google Earth's PhotoOverlay manually is a real pain. Now, if only Microsoft will do it in such a way that the photos can be output through KML using the PhotoOverlay parameter - that would be cool! Vincent also briefly demoed a "StreetView" technology they are working on - it was showing "video" of the street views. Another last minute demo Vincent gave was showing the ability to add sunlight time animations in Virtual Earth 3D in some future update. Microsoft is definitely pushing hard to earn the "best" title in virtual world technology. And, they certainly do have some of the best data (BEV, their ortho photos, the size of their 3D city models, and the new 3D trees in their v2 city models). I certainly hope Google is working hard to add new and better models and features or they may lose their lead.
• Poly9 - Poly9 are the creators of FreeEarth - a free browser-based 3D virtual globe which was one of the first to support KML. They have added a number of features to Free Earth - including some not even in Google Earth (like sound). More recently they have announced a new commercial product to help emergency responders called "GeoAlert" (see press release). They claim the system can help save lives. They've teamed with another company which can automatically call people when an emergency (like a chemical fire) requires evacuations.
• Fortius One - Sean Gorman - Sean has a good grasp of the development of the GeoWeb. He gave a good summary of progress in that arena and described efforts by his company with their GeoCommons (announced at Where last year) which lets people upload data and get back interesting visualization maps. He said they ran into a bottleneck with their design oncce they reached 1.6 billion uploaded features. They formally announced a new application for GeoCommons called Finder! whose mission is to allow you to "Upload, Organize, and Share your GeoData".
• Pict Earth and DIYDrones - Pict Earth's David Riallant and Jeff Johnson gave a great overview of their aerial photography technologies using primarily unmanned aerial vehicles (UAVs), and some regular general aviation aircraft. (I've written about Pict Earth several times because they have used Google Earth as a platform for visualizing their data). The significance of what they are doing is the real-time geo-location and mapping of the photos, and the experiments they have made with using off-the-shelf technlologies like smart phones to acquire the photos, provide GPS location, and communicate with the ground. They have also worked with Chris Anderson who has been developing technologies to create auto-pilot capabilities to turn basic remote controlled aircrafts into UAVs. Chris gave a very engaging talk also describing the evoloution of their technologies (and some great stories about succeses and failures). Their goal is to create open source software and very low cost solutions that anyone can add to create their own UAVs. See his web site DIYDrones.com.

I'll try to update this post with other observations from Day 2 later. But, I need to go prepare for today's sessions. I also want to mention a few things I observed in the booths and at the Where Fair.

Many Visualizations of Myanmar Disaster for Google Earth

I've been updating my original post with a list of visualizations on the Myanmar disaster for Google Earth showing Cyclone Nargis's path, weather data, and the resulting flooding from satellite photos. Google and other organizations have released all kinds of data in the aftermath of the storm to help people realize the extent of the disaster.

Google has a link for places to send support to Myanmar in their time of need.