FisheyeWarping 1.0.0

Dewarp top-down fisheye image to panorama and rewarp panorama to fisheye image.

Fisheye Warping

A tool for dewarping and rewarping on a top-down fisheye image by using OpenCV on Python.

Installation

pip install FisheyeWarping

Usage

fisheyewarping --help

usage: fisheyewarping [-h] [--panorama_output PANORAMA_OUTPUT] [--fisheye_output FISHEYE_OUTPUT] [--save_dewarp_mesh_path SAVE_DEWARP_MESH_PATH] [--save_rewarp_mesh_path SAVE_REWARP_MESH_PATH] [--load_dewarp_mesh_path LOAD_DEWARP_MESH_PATH]
                      [--load_rewarp_mesh_path LOAD_REWARP_MESH_PATH] [--fisheye_img_path FISHEYE_IMG_PATH] [--panorama_img_path PANORAMA_IMG_PATH] [--use_multiprocessing USE_MULTIPROCESSING]

optional arguments:
  -h, --help            show this help message and exit
  --panorama_output PANORAMA_OUTPUT
                        Specific path for `output`. Default is `./dewarp-output.png`.
  --fisheye_output FISHEYE_OUTPUT
                        Specific path for `output`. Default is `./rewarp-output.png`.
  --save_dewarp_mesh_path SAVE_DEWARP_MESH_PATH
                        Specific path for saving mesh data for `dewarping`. Default is `None`.
  --save_rewarp_mesh_path SAVE_REWARP_MESH_PATH
                        Specific path for saving mesh data for `rewarping`. Default is `None`.
  --load_dewarp_mesh_path LOAD_DEWARP_MESH_PATH
                        Specific path for loading mesh data for `dewarping`. Default is `None`.
  --load_rewarp_mesh_path LOAD_REWARP_MESH_PATH
                        Specific path for loading mesh data for `rewarping`. Default is `None`.
  --fisheye_img_path FISHEYE_IMG_PATH
                        Specific path of your fisheye image for dewarping to a panorama.
  --panorama_img_path PANORAMA_IMG_PATH
                        Specific path of your panorama image for rewarping to a fisheye image.
  --use_multiprocessing USE_MULTIPROCESSING
                        Use multiprocessing to get mesh. Default is `True`.

Example

Dewarp a fisheye image to panorama

• In the first time, you will need to build a mesh file for dewarping.

  fisheyewarping \
  --save_dewarp_mesh_path ./dewarp-mesh.pkl \
  --fisheye_img_path ./test-fisheye.jpg
  

  import cv2
  from fisheyewarping import FisheyeWarping
  fisheye_img = cv2.imread('./test-fisheye.jpg')
  frd = FisheyeWarping(fisheye_img, use_multiprocessing=True)
  frd.build_dewarp_mesh(save_path='./dewarp-mesh.pkl')
  frd.run_dewarp(save_path='./dewarp-output.png')
  

• In the next time, you just load the mesh and feel free to go.

  fisheyewarping \
  --load_dewarp_mesh_path ./dewarp-mesh.pkl \
  --fisheye_img_path ./test-fisheye.jpg
  

  import cv2
  from fisheyewarping import FisheyeWarping
  fisheye_img = cv2.imread('./test-fisheye.jpg'.)
  frd = FisheyeWarping(fisheye_img, use_multiprocessing=True)
  frd.load_dewarp_mesh(save_path='./dewarp-mesh.pkl')
  frd.run_dewarp(save_path='./dewarp-output.png')
  

Rewarp any panorama image to fisheye

• In the first time, you will need to build 2 mesh files for dewarping and rewarping by using one fisheye image.

  fisheyewarping \
  --save_dewarp_mesh_path ./dewarp-mesh.pkl \
  --save_rewarp_mesh_path ./rewarp-mesh.pkl \
  --fisheye_img_path ./test-fisheye.jpg
  

  import cv2
  from fisheyewarping import FisheyeWarping
  fisheye_img = cv2.imread('./test-fisheye.jpg'.)
  frd = FisheyeWarping(fisheye_img, use_multiprocessing=True)
  frd.build_dewarp_mesh(save_path='./dewarp-mesh.pkl')
  frd.build_rewarp_mesh(save_path='./rewarp-mesh.pkl')
  panorama_img = cv2.imread('./test-panorama.jpg'.)
  frd.run_rewarp_with_mesh(panorama_img, save_path='./rewarp-output.png')
  

• In the next time, you just load the meshes and feel free to go.

  fisheyewarping \
  --load_dewarp_mesh_path ./dewarp-mesh.pkl \
  --load_rewarp_mesh_path ./rewarp-mesh.pkl \
  --panorame_img_path ./test-panorama.jpg
  

  import cv2
  from fisheyewarping import FisheyeWarping
  frd = FisheyeWarping(None, use_multiprocessing=True)
  frd.load_dewarp_mesh(save_path='./dewarp-mesh.pkl')
  frd.load_rewarp_mesh(save_path='./rewarp-mesh.pkl')
  panorama_img = cv2.imread('./test-panorama.jpg'.)
  frd.run_rewarp_with_mesh(panorama_img, save_path='./rewarp-output.png')
  

Mesh

The source of mesh image is from (http://paulbourke.net/dome/fish2/).

Here is the mesh image for this tool, it shows the transformation from the original fisheye image to the panorama image.

• Before - Fisheye

• After - Panorama

Contributor

Nat Lee

LICENSE

MIT