IPM

1. GetIPMImage.py

• Main file to run
• Where parameters currently need to be adjusted manually
  • Focal Length (unit: pixel)
  • Camera Height
  • Pitch
  • Yaw
  • Roll
  • left, right, top, bottom (the road range in the photo that needs to be used for IPM)
• The current resolution of OutImage is tentatively 640*640. If the aspect ratio of the top view is wrong, you can manually adjust the length and width of the output image.

2. GetVanishingPoint.py

• It is used to find the vanishing point of the road (more suitable for photos with road ends). The purpose is to automatically find the top position, but I currently feel that the calculated results are not good, and I am trying new methods.

3. TransformImage2Ground.py

• is used to find the relative positions of left, right, top, and bottom in world coordinates

4. TransformGround2Image.py

• is used to find where each point in the original image that will be converted into a top view is located for coloring.

5. Example

• road1

• road2

• road3

Reference

1. IPM
   • https://blog.csdn.net/monk1992/article/details/83780999
   • https://blog.csdn.net/yeyang911/article/details/51915348
   • https://sites.google.com/site/yorkyuhuang/home/research/computer-vision-augmented-reality/ipm
2. AutoStitch64
   • http://matthewalunbrown.com/autostitch/autostitch.html