Light field as an enabler of multi‑view surveillance and photographic VR

Recent reports show that around 80% of internet traffic carries visual information which has benefited society in different parts of life such as education, medical, corporate, social, leisure, and sports. This sheer dominance of visual information over the internet is mainly due to advances in capturing, processing, and display technologies along with the continuous development of novel applications such as YouTube, Facebook, WhatsApp, Instagram, and Zoom.

In the pursuit of accurately capturing scene visual information, light field (LF) acquisition technologies were developed to record the spatial and angular information of the scene. LF data facilitates numerous post-processing applications, such as novel view generation, 3D scene reconstruction, and refocusing at the cost of increased data size. In the recent past, significant research activities have been carried out to improve the rate-distortion (RD) efficiency of LF data, which is an essential step to make its use possible in future applications. However, such improvement in RD efficiency comes at the cost of increased computational complexity. The proposed project light field as an enabler of multi-view surveillance and photographic VR (LFMV) investigates this crucial aspect: reducing the encoding time for LF contents while maintaining the high RD efficiency, which is essential to enable LF contents to be part of the transmission chain for future applications such as multi-view surveillance and virtual reality (VR) applications.

The first part of the project centers on academic research in light field compression with real-time requirements and the second part of the project focus on the industrial interest in multi-view surveillance and VR systems.