Johannes Turinsky, Michael Rieck
Few users of cheap webcams will be satisfied with the picture quality. Manufacturers often use cheap and poor lenses to lower the cost. Pictures taken with such a camera often have barrel distortions. Therefore we started to look for an image correction method that is cheaper than the use of better and more expensive lenses.
We decided to edit each pixel individually and transferred the RGB color values of every pixel of a webcam image into an Excel table. In order to get familiar with Excel and to keep the computational effort low, we started with "ideal" images of a black and white grid and a colored plane, which we created ourselves. We used these ideal images to simulate a distortion and calculated the difference between the ideal and distorted pictures. This way we identified the pixels that were not in the right place. We tackled the final step of moving these pixels to the correct locations by shifting the relevant fields in Excel using several functions.
In order to be able to correct distortions in any picture, it is important to remember how often each pixel was moved in which direction. After application of this key to distorted image data in our Excel table, the corrected table of color values needs to be converted into the corrected image file. It would be more user friendly to write a program based on our image correction algorithm. It would be best to integrate such a program into the webcam itself, so that corrected images are transferred to the computer in the first place.
