A PARAMETRIC VIDEO QUALITY MODEL BASED ON SOURCE AND NETWORK CHARACTERISTICS


Zerman E., Konuk B., Nur Yılmaz G., AKAR G.

IEEE International Conference on Image Processing (ICIP), Paris, Fransa, 27 - 30 Ekim 2014, ss.595-599 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/icip.2014.7025119
  • Basıldığı Şehir: Paris
  • Basıldığı Ülke: Fransa
  • Sayfa Sayıları: ss.595-599
  • Anahtar Kelimeler: Quality of experience (QoE), no-reference metric, video quality assessment (VQA), network condition, video characteristics
  • TED Üniversitesi Adresli: Hayır

Özet

The increasing demand for streaming video raises the need for flexible and easily implemented Video Quality Assessment (VQA) metrics. Although there are different VQA metrics, most of these are either Full-Reference (FR) or Reduced-Reference (RR). Both FR and RR metrics bring challenges for on-the-fly multimedia systems due to the necessity of additional network traffic for reference data. No-eference (NR) video metrics, on the other hand, as the name suggests, are much more flexible for user-end applications. This introduces a need for robust and efficient NR VQA metrics. In this paper, an NR VQA metric considering spatiotemporal information, bit rate, and packet loss rate characteristics of a video content is proposed. The proposed metric is evaluated on EPFL-PoliMI dataset, which includes different video content characteristics. The experimental results show that the proposed metric is a robust and accurate NR VQA metric towards diverse video content characteristics.