Who do you see in the set of images below? The famous physicist Albert Einstein? Or the pop culture icon Marilyn Monroe?
Figure 1: Einstein-Monroe hybrid image
You will be able to distinguish the physicist’s face when looking at the image at close range; and the cultural icon, from afar. This is a well-known optical illusion.This hybrid image overlaying high-frequency image and low-frequency images is also a typical example of a common image processing technique used to illustrate the ‘Einstein-Monroe illusion.’
There are several methods to achieve this effect, but we will continue to focus on Fourier transform for the purposes of this article.
As portrayed above, the farther perspective image of Marilyn Monroe undergoes Fourier transform with a low-pass filter; and near perspective image of Albert Einstein, a high-pass filter. When we perform 2D inverse Fourier transform(iFFT2) on the weighted images, we achieve the optical illusion effect of seeing two images different on perspective.Yet there is one important consideration: the low-pass and high-pass filtered images must uniformly coincide for the most unaffected composition.
Application Case Study
Thermal cameras are commonly used in manufacturing for various purposes, such as detecting equipment anomalies or determining product defects. Although infrared imaging can provide a visual representation of the temperature distribution of a target object, the low resolution and limited color scheme makes it difficult to determine a precise delineation of the target object.
