IEEE ICIP 2022

Carlos Cortés and Narciso García, members of the Grupo de Tratamiento de Imágenes, attended the IEEE ICIP 2022 Conference held in Bordeaux from 16 to 20 October.

" I had a pretty big audience, although it was one of the few papers that did not deal with neural networks." - Carlos said.

The goal of the presented work was to find the limits of self-view delay in immersive XR environments in terms of quality of experience. For this purpose, Carlos adapted for virtual reality a task from the ITU-T P.920 recommendation that consists in building figures using lego-like blocks. During the experiment subjects had to build specific figures while the delay varied for each of them. After each construction, the users had to complete a questionnaire previously validated for immersive interactive environments. The results of the experiment showed that the limit of acceptance of the self-view delay for presence and global perception of quality was above 450 ms (including the camera delay). On the other hand, users needed more severe delays to notice their effect on adaptive ability.

You can find the link to the conference paper here: https://ieeexplore.ieee.org/document/9897983

Carlos also recounted his experience at the conference: "During the conference, I was able to attend numerous presentations on the current state of the art in image processing methods. Obviously, most of them were based on deep learning techniques. However, I observed that many methods adapted their problems domain to the image one. The purpose of this is to use the existing neural networks for image processing because they are very robust and developed. In addition, other contributions focused on cleaning or selecting only a part of the input information to increase the performance of the networks. In addition, I attended a tutorial on point cloud compression where I learned a lot about point cloud compression using different strategies."

Tweet