Seminar 8 -S3, Tuesday 6th December 2022, 14:00 (London Time)

Speaker: Ronald Aznavourian (Institut Fresnel, Marseille, France)

Access link to live webinar

Title: Denoising of radar images by AI.

Abstract: A lot of applications are possible with AI: classification, to sort a lot of images in many different categories, denoising, to improve the quality of a signal or an image, and “generative”, to create a fake image which seems to be very real. Here, I propose to use an autoencoder to denoise radar images. In order to achieve this, I first propose to create an adequate dataset of 60,000 images to train the autoencoder.

Biography: Ronald’s background is mainly in computer science. He graduated from an engineering school in networks, systems and multimedia and worked on morphing applied to numerical simulations during his engineering thesis. He is now a PhD student, in the third year of his thesis, in the field of image processing, under the supervision of Julien Marot, at the Institut Fresnel. He uses AI, and more particularly neural networks to denoise images, as well as bio-inspired single-objective optimization methods, such as the Grey Wolf Optimizer (GWO), or multi-objective optimization methods based on Genetic Algorithms, such as the Non-dominated Sorted Genetic Algorithm-II (NSGA-II) to optimize neural networks or even invisibility cloaks.

Access link to live webinar