79 | Image Processing | Physical engineering and embedded systems | S9 | ||||||
---|---|---|---|---|---|---|---|---|---|
Lessons : 4 h | TD : 0 h | TP : 24 h | Project : 0 h | Total : 28 h | |||||
Co-ordinator : Loic SIMON |
Prerequisite | |
---|---|
Signal processing, Algorithms |
|
Course Objectives | |
1) Expose the classical modeling of images and acquisition noise as stochastic processes 2) Introduce standard principles of classical algorithms (filtering, inverse problems, …) |
|
Syllabus | |
Part 1: modelling - Image capture chain (simplified model of CCDs, transfer function, Bayer pattern, natural vignetting) - stochastic processes (reminders on random variables, modelling of distributions, stochastic processes, invariance) - acquisition noise (internal / external noise, shot-noise, quantification, noise, other noise models) - color spaces (RGB, CMYK, HSV, HSL) Part 2: algorithms - functional and technical specifications (task goals, constraints, complexity) - examples (integral images, histogram equalization) - linear filtering (taxonomy, FFT, application to contour extraction and sharpening) - Image restoration (denoising, non-blind deconvolution) - Inverse problems (formulation, gradient descent, handling of outliers) - Interpolation - Short introduction to deep learning |
|
Practical work (TD or TP) | |
1) Basic image processing operators in Matlab 2) Linear filtering and FFT: application to image restauration 3) Image segmentation 4) Case studies: medical imaging; shape recognition; |
|
Acquired skills | |
- General skills (stochastic modelling, algorithmic notions, optimization) - specific skills (image capture chain, image processing standard formulations, efficient solutions) |
|
Bibliography | |
Gonzales, Rafael C., and Richard E. Woods. "Digital image processing." (2002). |
© 2024 - ENSICAEN ( Legal Notices - Credits )