Friday, April 13, 2012

Problems With Reconstructing Pinhole Images

It can be difficult to see fine details in a B&W image and another way of viewing the simulated pinhole image is by looking at the changes instead. In the images below the changes between neighboring pixels were computed and stretch in the grey scale to fit between 0 and 255. The zero point was chosen to be 50% grey. White indicates a positive difference and black a negative difference.

The two contrast images show how the small dark point at the center of the test object affects the pinhole image. There is a slight decrease when the aperture includes these points.

I also noticed a lot of round off errors in the aperture sums. The sums ranged from 0 to 317 (the number of aperture pixels) which were adjusted to fit between 0 and 255 for the image. Converting back to the original range produced the round off errors of ±1.

The errors occurred where the changes took place mainly within a range of ro±ra from the center of the disk where ro was the radius of the disk and ra was the radius of the aperture. The errors make it more difficult to reconstruct an image.

No comments: