If you want to stay home all over the world, then Google’s Street View service is definitely a good choice. However, these images are collected through the car, Trekker backpack, and diving equipment. But in the acquisition process will inevitably shift a variety of photos and other issues, the natural person responsible for the acquisition of the capture cannot go to the scene to re-shoot once, then How does Google fix these photo issues?
Google made some effort to solve the problem, recently Google Maps use Machine Algorithm to fix street view image problems, which looks very effective.
Machine Perception, a research scientist from Google, and senior engineer Mike Krainin, a software engineer, showed the secrets behind it.
Krainin said that in order to provide a more seamless street view images, we are addressing these challenges based on a new algorithm for light flow, which creatively distorts each of the input images and rearranges the overlapping parts.
After many iterations of the algorithm, Google finally created the ideal algorithm to solve the problem. The computer analyzes segmented areas of Google Street View images and then makes overlapping adjustments to create a complete image in the human eye.