Problem 1 [70 pts]: In this problem, you will perform camera pose estimation using homography. Given this video your task is to compute the rotation and translation between the camera and a coordinate frame whose origin is located on any one corner of the sheet of paper. In order to do so, you must:
- Design an image processing pipeline to extract the paper on the ground and then extract all of its corners using the Hough Transformation technique .
- Once you have all the corner points, you will have to compute homography between real world points and pixel coordinates of the corners. You must write your own function to compute homography.
- Decompose the obtained homography matrix to get the rotation and translation
- Note: If you decide to resize the image frames, you need to accordingly modify your intrinsic matrix too. Refer to this discussion.
- Data: The dimensions of the paper is 21.6 cm x 27.9 cm. The intrinsic matrix of the camera can be found here.
This projects consists of the following code files
- Problem #1:
- Code: Problem1.py
- Data: project2.avi, K_matrix
- Results (video marked with lines, and corners, pose estimatin plots)
- None
-
Ensure the following depenancies are installed
pip install numpy pip install scipy pip install matplotlib pip install opencv-python -
Ensure that the above programs are downloaded into the same folder containing 'project2.avi'
- Problem1:
- Run the program "Problem1.py" to check the outputs.
- As the program gets executed, the video, with detected lines and corners, can be seen.
- The program will run for 1-2 mins, after which the camera pose estimation plots, number of lines plots are displayed.
- The file - "HomographY_check", also gets generated. This is a text file to cross-check the generated matrix.
-
- On running each of the proframs, the output either pops out a plot or a video in individual window. The outputs can be correlated with the outputs shown in the report.
