The comparison and improvement of data processing methods in video quality subjective assessment

Data screening applied in video quality subjective assessment is helpful to remove abnormal observers and ensure the precision and effectiveness of the results. This paper firstly introduces two main data-processing methods, which are recommended in ITU-R BT.500 and ITU-R BT.1788 respectively, and then proposes an improved data-processing method, which is used to eliminate observers whose score trends are inconsistent with the overall trend in the first stage and then eliminate observers with too many outliers. Finally we analyze the applicability of those data-processing methods and verify the effectiveness of the improved method.