Google’s Square Street View Pattern

I’m sure if you’ve used Google’s Street view, you will have noticed that the images are mostly in larger metro areas. If you zoom out from those areas, you can see that the image data in confined to neat “squares” around those metro areas.

Picture 1-1

I though this was an off way to organize this information. I mean it’s quite obvious that Google cannot photograph EVERY street in the world right away, and it makes sense to start with highly populated areas first. However the notion of squaring off these ares seems to be sort of counter intuitive. Because as I will show, the is most likely data being collected by Street View cars that is not being shown.

Just some FYI, Google uses cars with cameras that take a series of photographs, and then those photographs are stitched together to form the pan-able view you have in Street View. Rather and taking a view of the vehicle in motion, Google Cars take one of these omnidirectional images every so many feet.

(Google seems to be in a hurry to finnish this project :) )

I’m sure that there is some short of cohesive master plan for the paths of these cars. But in the end all the drives really have to do is to wonder down every road in America, and looking at the street view maps, it really does seem like “wonder” is the appropriate word.

Have a look at the Knoxville, TN Street View availably.

Picture 2-1

Here the South-East and North-West corners of this square are sparse. This is because there are few if any roads in those areas. If we have a closure look at the North-West corner, we can see where the Google Car went down some side roads, but in down so seemed to cross the invisible line that marks the top of this square.

Picture 4-1
Here is the live data view from Google

It’s pretty obvious that the Google Car Followed Stoney Fork Rd through Sheam, then made a right onto Clinchmore Rd before stopping. What’s odd about this is that all the data from north of this inviable line is not on Google Street View.

This posses a couple of questions. Do the cars stop collecting data when they pass outside of the defined box? This is unlikely, because in the is example, if Google expanded the Square, then the cars would has to retrace this part of the road. So in all probability, Google Cars are collecting the data, but it is not being displayed. Why Not? It understandable that it takes time to Process a Street View section of Map. The Car Camera images have to be stitched together, the routines setup in Google Maps, etc. But this just begs the question, did Google just not process the data north of this line? This also seems unlikely, given that again in the case the car drove south of the line at two points. So why skip over data to process if it was coming off the car in a linear fashion.

In all likely hood there is probably some sophisticated geo-taging going on with each set of images these cars take. I would not be surprised if the omnidirectional images are not assembled on-the-fly by computers in each Car as they are taken, the the geo-taged with high resolution GPS. These data blobs can be dumped like a hand full of marble into the Street View databse. Software at Googles end can then assemble the “chains” and “connection points” each of these blobs have with other blobs based on the geo-data. I’ve not done the research to see just how google is doing this, but this seems the logical away to put it together.

That just leave the question of the inviable line. If it’s likely that Google Cars took the pictures, and those pictures have been processed, then why is the data not up? Google seems to be cutting it off at an arbitrary point. Also, why use Squares? Circles seem to be easier to do than squares. If Google really is artificially suppressing some of the street view data, why not do it using a Circle vector. With a square you have to define four points (or four lines), basically 8 units of data. A Circle is can be defined by a single point and a distance factor, 3 units of data. As you are ready to replace more street view data, Google simple ups the distance factor to include more space.

Finally, the most convincing part of the artificially suppressing argument, is the lack of Interstate views. It’s unlilky that Google Cars are based in every metro area in the us. They seem to drive around. So, do they have the cameras turned off while on the Interstates?

If you look at the separation of the squares between Dalton Ga, and Adairsville, GA, which is only about 8 miles, you can’t tell me that bit of I75 has not been photographed or processed.

Picture 3-1

It’s also clear that GA136 was traveled by a Google Car because there’s yet another orphaned bit of road. Why is GA136 between Vihanow and Sugar Valley not on Street View?

Perhaps Google has a reason for doing this, I’m just cannot seem to understand why.

Leave a Reply