Marshall University Math Colloquium
April 2, 2012
Kansas State University
The central problem in our presentation is the following: Suppose we have an image and we want to find the best match to this reference image from a large database of images. The best way to tackle this problem is indeed to let human observers, the ultimate receivers of all image information, be the judge. However, due to the tedious and time-consuming nature of this problem, we would rather employ a computer algorithm which mimics the human visual system in recognizing similarity between images. In this presentation, we will describe some of the challenges of this problem and discuss some image quality assessment methods which attempt to address these challenges. In particular, we will see how tools from Statistics and Fourier Analysis can come into play in the development of such methods.