Comparative Study of Data-Adaptive Structure Tensors

Dec 12, 2018 - Context: Thin structures in images are ubiquitous. They are usually characterized by a smaller size in at least one of their dimensions. Standard ...
387KB taille 2 téléchargements 353 vues
Comparative Study of Data-Adaptive Structure Tensors December 12, 2018

Context: Thin structures in images are ubiquitous. They are usually characterized by a smaller size in at least one of their dimensions. Standard regularization operators (such as Total Variation) used in the variational framework are not adapted to preserve such structures in image restoration or image segmentation. In some medical applications, their preservation can however be crucial. To overcome this problem, a number of approaches have been proposed along the past years [1, 2]. In [1], the benet of using data-adaptive structure tensors is demonstrated on typical vision tasks (see Fig 1). However, the comparative study is limited on few tensors and data. This study could be for instance extended with more recent and promising works using Bilateral lters [3]. Objectives: First, a thorough bibliography will be conducted on the problem, including the most recent approaches. Second, a reasonable subset of these approaches will be implemented. Finally, a comparative study will be led on both simulated and realistic images in image segmentation and/or image restoration. Prerequisites: A highly motivated candidate is expected at master level (or equivalent) with excellent mathematical and image processing background as well as good programming skills and technical english level. Knowledges in optimization are preferred but not mandatory. Dates/salary: Approximately from February to July / About 530 euros per month. Location: The internship will take place in the SATIE lab at Gif-sur-yvette (30 minutes from Paris). Contact: Please feel free to send an e-mail to [email protected].

References [1] T. Brox, R. van den Boomgaard, F. Lauze, J. van de Weijer, J. Weickert, P. Mrázek, and P. Kornprobst. Adaptive Structure Tensors and their Applications, pages 1747. Springer Berlin Heidelberg, 2006. [2] O. Merveille, O. Miraucourt, S. Salmon, N. Passat, and H. Talbot. A variational model for thin structure segmentation based on a directional regularization. In International Conference on Image Processing (ICIP), pages 43244328, 2016. [3] L. Zhang, L. Zhang, and D. Zhang. A multi-scale bilateral structure tensor based corner detector. In Asian Conference on Computer Vision, pages 618627, 2010.

Figure 1: Simulated image (left) and linear structure tensor (right). 1