Project -6

Department of Electrical Engineering and Computer Science

CIS/EEC 693

Computer Vision

Summer 2021

Project 6

(Due date: 06/21/2021)

The objective of this project is to perform 3D indoor scene reconstruction.

1. Write a program to create a 3D point cloud model based on the pinhole camera model. Use one RGB-D image (image1 rgb with image1 depth) that is captured using a Kinect camera,

see included Data5.

The essential parameters of RGB-D Microsoft Kinect Camera are as following:

• Focal length - fx = 521

- fy = 521

• Camera center - cx = 325

- cy = 250

2. Repeat part 1 using two RGB-D images (image1 rgb with image1 depth and image2 rgb with image2 depth) to estimate the camera position assuming the camera position of image1 is at

the origin (0,0,0).

• To estimate the camera transformation matrix (T), apply Iterative Closed Point (ICP) method using point-to-plane error metric (or you may use any built-in function).

• Once the T is estimated, transform point cloud of image 2 to image 1 camera coordinates (you use any built-in function).

• At the end, merge the two 3D point clouds.

Notes:

• The project should be implemented in Python.

• Only one single file should be submitted through Blackboard for evaluation, which includes:

✓ The project report that includes the methodology, equations used, implementation results and discussion, conclusion, appropriate technical references, etc.

✓ The program codes along with the dataset used for testing and validation.

• Late submissions will not be accepted. • Email submissions will not be accepted.