Institute for Theoretical Information Technology

Sampling Pattern on the Sphere and Performance Evaluation of Sparse Signal Recovery

Compressed Sensing (CS) is concerned with the recovery of sparse signals using only few samples and it has sparkled a significant amount of research after the pioneering works. Assuming that we are performing a measurement on the spherical surface, what is the best sampling pattern to recover or extract information of the signal?

In this project, we will implement several well known sampling patterns on the sphere and evaluate its performance to recover sparse signal.

Content:

1. Wigner-D Basis and Spherical Harmonics as a basis expansion

2. Evaluate several sampling pattern on spherical surface

3. Deterministic Samplin

• a. Spiral
• b. Fibonacci
• c. Platonic solid
• d. Healpix
• e. Hammersley
• f. Maximum determinant
• g. Minimum energy

4. Random Sampling

• a. Random uniformly distribution on a sphere
• b. Dense pole and equator distribution
• c. Von Mises-Fisher distribution

5. Evaluate the coherence of each sampling pattern

6. Recovery algorithm : BP, OMP, IHT

7. Phase transition diagram or MSE analysis

Extension:

• 1. Phase retrieval and superresolution on the sphere

Dates

The first meeting for the institute project is on Friday, 28.04.17, 3 p.m., in ICT cubes Room 333.

Contact Person: M.Sc. Arya Bangun upon agreemnet

zurück zur Liste der Vorlesungen