Depth estimation and segmentation from street view using digital map

Abstract

In the game, the background is not only a spatial element, but also an element that increases game immersion by giving the story a probability. For this reason, some games with a background modeled on an actual city center are gaining popularity, through games, players can experience places that they cannot visit in real life. It is difficult to create a game set in the city center because it requires a lot of manpower and time to produce background. On the other hand, there are also games that provide virtual experiences to users using services such as Street View and simple rules. It will be very useful if you can create a map based on reality and increase the user's immersion by utilizing the building that is an important element in the background based on Street View. Usually in games, depth maps are used for lighting and shadow generation, or constructing terrain like height maps. We are going to apply a depth map to street view. In this study, we intend to experiment with a simple and low-cost depth estimation algorithm using a digital map. The location and metadata of the building were acquired from the digital map and the depth was estimated. We added depth information to Street View to highlight important buildings as background elements and separate buildings, ground, and sky. Street view studies based on digital maps and metadata can be used in a variety of ways, such as building 3D maps and advertising based on building planes.