Impacto de Erros de Canal na Percepção de Qualidade em Sequências de Videoconferência de Alta Definição

In this work we assess the impact of channel errors on the perceived quality in high definition videoconference applications. A database containing 31 high definition sequences (1080i) without compression was generated. The set of generated sequences may present backgrounds of different complexities as well as other features, such as number of subjects, face occlusion, skin color, among others. 16 representative sequences were degraded with channel error simulating packet loss that ranged from 0 to 0.05% and post processed by error concealment techniques. Results suggest that the localized nature of channel errors (in contrast to encoding errors) may mask the perception of quality by the users in a way that are not associated to the error rate itself, but also to the nature of the video content.