1 / 7

Progress Report: Week 8 Alvaro Velasquez

Progress Report: Week 8 Alvaro Velasquez. Framework. Create data matrix of all non-overlapping 4x4x4 cuboids in image volume. Learn dictionary for the data matrix using KSVD. Obtain sparse coefficient matrix using graph regularization. Cluster coefficient matrix using K-means.

Télécharger la présentation

Progress Report: Week 8 Alvaro Velasquez

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Progress Report: Week 8 Alvaro Velasquez

  2. Framework • Create data matrix of all non-overlapping 4x4x4 cuboids in image volume. • Learn dictionary for the data matrix using KSVD. • Obtain sparse coefficient matrix using graph regularization. • Cluster coefficient matrix using K-means. • Display clusters on image volume for segmentation.

  3. Things Tried this Week • Append all three RGB channels to vectorized cuboids as opposed to using gray-scale • Use Pearson correlation for the distance measure when clustering as opposed to Euclidean distance. • Choose initial centroid locations when clustering as opposed to choosing random locations. • Use Max-Voting for initial centroid locations. • Perform Quick-shift as a preprocessing step.

  4. Last Week's Segmentation

  5. Choosing Initial Centroid Locations

  6. Centroid Locations Through Max-Voting

  7. Work for this Week • Add Gaussian smoothing as a pre-processing step. • Make sparsity rate larger. • Try different dictionary sizes (We have tried 1000 and 2000 atom dictionaries).

More Related