Take a look at Google Street View 来看一看谷歌街景

One man I followed on Pinterest posted one photo about what he met with Google. He tagged it as “Google-self-driving Car” like below, but actually, from the equipped sensors, this vehicle is not a self-driving car, but a street view car.

我在Pinterest上关注的一个外国朋友上传了一张他拍摄的Google汽车的照片.他将这张图标记为”谷歌无人驾驶汽车”,如下图所示.但实际上从该车的传感配置来说,这辆汽车并不是无人驾驶汽车,而是谷歌的街景汽车.

The street view, is a project combined with Google Earth and Google Map. As Google is independent on the high-detailed map, some of the technologies are shared between different labs in Google.

谷歌街景是和谷歌地球谷歌地图相互依赖的项目.由于谷歌一直依赖于高精度的地图,某些技术在谷歌的不同实验室之间是共享的.

Let’s take a look at the official image of Street View Car from Google.

我们来看看来自官方的街景汽车的照片.

Comparing the two above image, you can easily find that they share a lot of common in the configuration of sensors:

  1. Camera that captures 360 degrees
  2. Laser beams for measuring objects’ distances
  3. GPS/IMU for measuring position, speed, and direction.

对比上面两幅图像,你可以轻易地发现他们在传感器配置方面有很多一致的地方:

  1. 采集360度全景的相机
  2. 用于测量物体距离的激光
  3. 用于测量车的位置 速度和方向的GPS/IMU

The generation of Street View follows the four major steps: (from map.google.com)

  1. Collecting imagery: First off we need to actually drive around and photography the locations to show in Street View. We pay close attention to many factors, including the weather and the population density of various areas, to determine when and where can collect the best possible imagery.
  2. Aligning imagery: To match each image to its geographic location on the map, we combine signals from sensors on the car that measure GPS, speed and direction. This helps us reconstruct the car’s exact route, and even tilt and realign images as needed.
  3. Turing photos into 360-degree panoramas: To avoid gaps in the panoramas, adjacent cameras take slightly overlapping pictures, and then we “stitch” the photos together into a single 360-degree image. We then apply special image processing algorithms to lessens “seams” and create smooth transitions.
  4. Showing your the right image: How quickly the car’s three lasers reflect off surfaces tells us how far a building or object is, and enables us to construct 3D models. When you move to an area in the distance, the 3D model determines the best panorama to show you for that location.

谷歌街景的生成按照如下四大步骤进行:

  1. 采集图像:首先在需要显示街景的地方进行驾驶和图像采集.采集前需要对很多因素进行评估:包括天气,当地人口密度,以决定何时何处有利于采集最好的街景图像.
  2. 配准图像:为了保证每幅图像与相应的地理环境对准,我们融合了测量GPS,速度和朝向的传感数据.这些数据有利于我们复现车的行驶轨迹,甚至可以根据需要倾斜并重新调整图像
  3. 生成全景:为避免全景图中出现间隙,相邻摄像机拍摄的照片会留有一定重叠区,后期我们再将照片拼接在一起,生成360度的图像,并使用特殊的图像处理算法以减少"接缝"并实现平滑过度.
  4. 显示合适的图像:通过配备的激光可以测量出建筑物或者物体与我们之间的距离,并根据次构建出3D模型.当用户移动到远处区域时,此3D模型可确定要在该位置为用户呈现最佳的全景图.

Where is the street view? See below.  It’s interesting to know why China is not covered. I’ve to admit, as I’ve living in Italy for almost one year, I am quite used to using Google Maps and Google Street View for navigation during my traveling. I am a little worried about this when I am back to China.

哪里有街景?请看下图.要搞明白为什么中国不在覆盖范围内会有趣.我得承认,在意大利生活近一年后,我已经非常适应于使用谷歌地图和谷歌街景在旅游中做导航使用.对于回国后的这个问题有点小担忧.

One problem is: is this car suitable for collecting image in every place? Obviously not. That’s why Google provides a lot of funny Street View Trekker, Street View Trolley, Street View Snowmobile, Street View Trike. So one day when you see something strange like these in your  block, don’t be too surprised.

有一个问题:这辆街景车就适合在所有地方采集街景了吗?显然不是.因此谷歌提供了一系列有趣的街景Trekker, 街景Trolley和街景雪车.如果某天你在你的小区发现上述怪物,不要过于大惊小怪.

 

关于dixiaocui

无人驾驶汽车研究员 半吊子摇滚混子乐手 Researcher on Intelligent Vehicle Guitarist/Singer/Rocker
此条目发表在Autonomous Driving分类目录,贴了, , 标签。将固定链接加入收藏夹。

Take a look at Google Street View 来看一看谷歌街景》有2条回应

  1. 兔子说:

    有中文的确感觉更好哦~

发表评论

电子邮件地址不会被公开。

此站点使用Akismet来减少垃圾评论。了解我们如何处理您的评论数据