UE E9TS-B - Advanced image and video processing methods

Level of knowledge :

N1: beginner
N2: intermediate
N3: confirmed
N4: expert

The knowledge (skills) expected at the end of the course

Know the principles of segmentation in a processing chain: (C1, N2)
Know region segmentation methods (split/merge, Markov models, seed growth, watershed): (C1, N3)
Knowledge of boundary segmentation methods based on derivative operators (Sobel, Prewitt, Kirch, Mdif, Nagdif, Laplacian, Canny): (C1, N3)
Acquire notions of morphology (structuring element, basic or advanced operations, connectedness): (C1, N2)
Knowledge of labeling algorithms and their applications (granulometry, skeletonization, morphological filters, etc.): (C1, N2).): (C1, N2)
Know the pinhole camera model and the homographic model: (C1, N2)
Know the formalisms of rigid 3D transformations (rotation and translation of a camera in a 3D environment): (C1, N2)
Acquire notions of epipolar geometry (essential matrix and fundamental matrix): (C1, N3)

Learning outcomes in terms of abilities, skills and attitudes expected at the end of EU courses

Segment an image using the watershed technique: (C2, N3)
Implement labeling using a two-pass algorithm: (C2, N3)
Use computer vision APIs (point-of-interest detection, matching, etc.): (C2, N2).): (C2, N2)
Implement RANSAC, Bundle Adjustment and Semi-Global Matching algorithms: (C2, N3)
Implement a motion-based video sequence processing chain (optical flow calculation, dense motion field representation, segmentation of high-motion regions): (C5, N3)
Segment an image into regions using a statistical approach: (C2, N2)
Estimate the orientation of a region in a partition: (C2, N2)
Apply a rotation to an image using bilinear interpolation: (C2, N2)