5LSH0 (2017) - Advanced Video Content Analysis and Video Compression

Goal:
Learn the theory and algorithms of video signal analysis, and finding features and objects in image/video. Learn the basics of 3D image processing, sensing in 3D and 3D model reconstruction. Learn the practical basics of MATLAB programming and analysis and/or 3D algorithms for practical applications in surveillance and medical imaging and learn the basics of deep learning.

Contents:
Content of the course is divided globally into three areas.

  1. New transformation techniques (wavelets and projective) for analysis in medical domain and 3D modeling in surveillance. The application of wavelets is given using the JPEG-2000 standard.
  2. 3D processing based on the camera pinhole model, homography and multi-view processing and calibration. Also registration of 3D datasets, 3D reconstruction models with TSDF, introduction to SLAM, RGB-Depth processing and specific algorithms like G2.0 and bundle adjustment. Finally, the 3D processing modules end with plane/object segmentation in 3D.
  3. Techniques for object detection and recognition, feature extraction and analysis, like SIFT and Laplacian methods. Furthermore, semantic level processing for understanding events and scenes, including human behavior. Furthermore, classification techniques for understanding objects and events. Modern classification like K-means and SVM (support vector machine) algorithms, evolving into basics of learning with neural nets. This part will gradually evolve to deep learning fundamentals.

The computer assignments aim at applying the knowledge and algorithms (or parts of them) to provide the student a framework for experiments with video content understanding and 3D image-babsed modeling for both surveillance and medical applications.

Preknowledge:
5LSE0 - Multimedia Video Coding and Architectures (recommended)
5XSA0 - Introduction Medical Imaging Processing (recommended)

Schedule and location: 
All lectures in a single week at the end of August, about 40% is dedicated to computer exercises.
Full schedule: 
Date Time Room Content Lecturer
28 aug 09.30-12.30 Flux 1.11 Module 1A+1B: Wavelet transformation theory + JPEG-2000 Prof. P.H.N. de With
28 aug 13.30-14.30 Flux 1.11 Module 1B: JPEG-2000 Prof. P.H.N. de With
28 aug 14.30-17.30 Flux 1.11 Module 2: 3D Pinhole camera, projective transformation Dr. E. Bondarev
29 aug 09.30-12.30 Flux 1.11 Module 3: 3D Registration, data fusion and SLAM reconstruction Dr. E. Bondarev
29 aug 13.30-17.30 Flux 1.11 Module 4: Visual feature extraction Dr. S. Zinger
30 aug 09.30-12.30 Flux 1.11 Module 5: Motion analysis and estimation Prof. P.H.N. de With
30 aug 13.30-17.30 Flux 1.11 Module 6: Object-level content analysis - segmentation A. Pourtaherian MSc.
31 aug 09.30-12.30 Flux 1.11 Module 7: Object-level content analysis - tracking A. Pourtaherian MSc.
31 aug 13.30-17.30 Flux 1.11 Module 8: Semantic-level content analysis and clustering + classification F. van der Sommen Msc.
1 sept 09.30-12.30 Flux 1.11 Module 9A: Deep learning introduction F. van der Sommen Msc.
1 sept 13.30-16.30 Flux 1.11 Module 9B: Deep learning work out F. Ghazvinian Zanjani
1 sept 16.30-17.30 Flux 1.11 Module 10: Case study application Prof. P.H.N. de With
Slides: 
Code of conduct
Module 1A: Wavelets
Module 1B: JPEG2000
Module 2: 3D Pinhole camera and projective transformation
Module 3: 3D Registration, data fusion and SLAM reconstruction
Module 4: Visual feature extraction
Module 5: Block Affine Motion
Module 6: Object-level content analysis - segmentation
Module 7: Object-level content analysis - tracking
Module 8: Semantic-level content analysis and clustering + classification
Module 9A: Deep learning introduction
Module 9B: Deep learning work out