A Smoother Street View

New street-level imaging software developed by Microsoft could help people find locations more quickly on the Web. The software could also leave new space for online advertising.

Services like Google Street View and Bing Streetside instantly teleport Web surfers to any street corner from Tucson to Tokyo. However, the panoramic photos these services offer provide only a limited perspective. You can’t travel smoothly down a street. Instead, you have to jump from one panoramic “bubble” to the next–not the ideal way to identify a specific address or explore a new neighborhood.

Microsoft researchers have come up with a refinement to Bing Streetside called Street Slide. It combines slices from multiple panoramas captured along a stretch of road into one continuous view. This can be viewed from a distance, or “smooth scrolled” sideways (see video).

“Today’s services plunk you down inside a bubble in a particular location,” says Michael Cohen, a senior scientist at Microsoft Research. “[Street Slide] helps you actually navigate using street- side imagery.” Cohen developed Street Slide with colleagues at Microsoft; the group’s work will be presented at the SIGGRAPH 2010 computer graphics conference in Los Angeles later this week.

A person using the original version of Streetside (as well as Google’s Street View) can rotate–within a “bubble”–to look in any direction. But this provides only a limited view of the buildings on a street, explains Cohen. “You want to back up until you can see the whole street but can’t because the buildings on the other side are in the way; we create that viewpoint using images taken from all the bubble panoramas along a street.”

Someone using Street Slide’s panoramic view can slide along the facades looking for places of interest, and zoom back in to a classic bubble view at any time. A user can also flip the viewpoint to see the other side of the street, or turn corners onto new streets.

The wider view provided by Street Slide offers empty space on the screen below the image of the street. This space could be used to display the logos of businesses, as well as a small map of the area. The space could also be used for advertising, Cohen says, or to display social information, such as the location of friends–if linked up to a social network.

Cohen and colleagues asked 20 people to find a variety of places on unfamiliar streets using Street Slide and Google Street View. Street Slide proved significantly faster–by 17 seconds, on average.

“This can really make an impact on a mobile device,” says Cohen, whose group has already made a version of Street Slide compatible with the iPhone. “It broadens out your visual sense to cover a two-block radius.” For example, he says, a person in an unfamiliar city could rapidly scan nearby streets for a café worth visiting.

Being able to visually scan a street is particularly useful in built-up areas, where interference from buildings can limit the accuracy of GPS location fixes, says Cohen.

Oh agrees, but points out that improved image-based mapping would benefit from better GPS location. “Geolocation accuracy was not that important when people were looking only at maps, but when you go down to the level of the street, you need a lot more [precise] information,” he says. Today, the location of many businesses–in the geolocation databases used by navigation devices and online maps–are estimated by uniformly spreading street numbers along a block.

EveryScape currently offers panoramic imagery from inside some 1,300 Boston-area restaurants via an app developed using the Bing Maps software development kit. “For seamless integration with street-side imagery, we want to link our interior images to the entrance point, the door–not just roughly where the premises are,” says Oh.