The task was to stabilize an old movie so that the frames did not move around and also to correct or normalize the gray scale of the image. We solved this using Matlab and explained in a simple way: looking at a characteristic part of the image and then compare that part to the next frame we were able to stabilize the movie clip. By looking at the histogram of the images we also could correct the gray scale. This program was created in a general way to work on any source.
Showing posts with label Matlab. Show all posts
Showing posts with label Matlab. Show all posts
Thursday, December 13, 2007
Movie restoration
A tool for restoration of old movies.

The task was to stabilize an old movie so that the frames did not move around and also to correct or normalize the gray scale of the image. We solved this using Matlab and explained in a simple way: looking at a characteristic part of the image and then compare that part to the next frame we were able to stabilize the movie clip. By looking at the histogram of the images we also could correct the gray scale. This program was created in a general way to work on any source.
The task was to stabilize an old movie so that the frames did not move around and also to correct or normalize the gray scale of the image. We solved this using Matlab and explained in a simple way: looking at a characteristic part of the image and then compare that part to the next frame we were able to stabilize the movie clip. By looking at the histogram of the images we also could correct the gray scale. This program was created in a general way to work on any source.
Red eye removal
A simple red eye removal tool using Matlab.
It seems like a simple task, but red eye removal is quite complicated. It includes locating the eye and removing the redness and make the eye look natural with a believable color and specular shine. This was a task that was given to us in the Advances Image Processing course. We completed the program using matlab and got a really good understanding of the problem.
A few persons have asked me if I could publish the matlab code, but I wont do that. Many of you are in school projects and should solve the problem yourself. This seems to be a trivial problem but it is in fact quite complicated. There are many, many ways to solve this and I can give you some hints to a good solution. Do not try to do an automatic solution that scans the whole image, you will then need to use som face-recognition algorithm and you do not want to go there, instead use the built in functions in matlab to mark an area. Then search for round areas in the red channel and do some opening/closing to get rid of noise. When you are changing the color, be sure that you dont remove the specular spot in the eyes becaus that will kill the person in the photo :) (It wont look realistic).
Hybri
A simple hybrid halftoning method written in Matlab.

I wanted to see how different halftoning methods worked together and how different areas or characteristics in the image could be used for choosing halftoning method in that area. A rasterization method can have good and bad properties. Then, what about combining the good properties of different halftoning methods and use them in the same image. To get an overview of this I created a simple GUI in Matlab where the user can select halftoning methods for different areas in an image.
Download matlab source (6 MB)
Download report (swe) (0.5 MB)
I wanted to see how different halftoning methods worked together and how different areas or characteristics in the image could be used for choosing halftoning method in that area. A rasterization method can have good and bad properties. Then, what about combining the good properties of different halftoning methods and use them in the same image. To get an overview of this I created a simple GUI in Matlab where the user can select halftoning methods for different areas in an image.
Download matlab source (6 MB)
Download report (swe) (0.5 MB)
Subscribe to:
Posts (Atom)
