Skip to content

IMAGE COMMUNICATION - EPFL COURSE June 2012 Inpaiting of depth image Yannik Messerli and Nicolas Jorns Supervised by Thomas Maugey

Notifications You must be signed in to change notification settings

yannikmesserli/Depth_inpainting

Repository files navigation

Inpaiting of depth image

IMAGE COMMUNICATION - EPFL COURSE June 2012

Author

Introduction

When estimating a camera view point from a first camera, we offen use a depth map for a better scene understanding. Naturally occlusions occur in both depth map and texture.

You'll find Matlab code lines produced during the mini-project during the course of Image communication done in EPFL. There is a lot of different tests and no final algorithm has been written but there is a lot of comments inside the files.

Files

The project is separated into different part with different tests

Estimation of the view

The first thing is to estimate the view from the second camera. Everything is inside the 'estimation' folder

Search for the different plane in the image

Then we must find the different plane around the holes. There is a lot of different tests for this part. Some comments and final tests can be found in the folder 'search for plane'.

Find and split the region

Then we must split the hole with respect of the regions. It's 'multi-plan' folder.

Test on specific image

'test_ball' is a artificial test on an image.

'test_baller' is a test on real depth images.

About

IMAGE COMMUNICATION - EPFL COURSE June 2012 Inpaiting of depth image Yannik Messerli and Nicolas Jorns Supervised by Thomas Maugey

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published