| Abstract: |
| In this talk, we will review the latest developments regarding using the eigenvalues of linear operators for shape recognition. In particular, we will discuss what makes the properties that make the eigenvalues desirable and reliable feature vectors. We will also present the image data sets and the recognition rates obtained on them for different operators. In addition, we will present some numerical results showing evidence that there might be a one-to-one correspondence between the eigenvalues of operators and moment invariants which are standard tools used for shape recognition. |
|