VQEG高清电视测试计划.ppt

* Step 1: Subjective testing Select sequences and run a subjective test to get MOS results. Include as many display technologies as possible. Make sure a variety of HRCs are used for the test sequences. Compile results from multiple labs to get a consolidated score for the sequences. * Step 2: Evaluation of existing models Run through existing (already submitted) models through the same sequences. Follow procedure and rules for calibration in the test plan. Include full reference, reduced reference and no reference models in the test. * Step 3: Results analysis and next steps Analyze objective results to get a correlation to the subjective results. Follow evaluation metrics in the test plan. Decide on next steps based on the results. If correlations are high, we may have a model for HD TV. If correlations are low, proponents will have a test set to tweak their models. * Lab readiness poll Of the labs represented here: How many of them have HD playback capability? Of those who don’t have it now, how many of them are planning to have this capability in the next 3 - 6 months. How many labs are willing to run HD subjective tests in the next 3 months? * Outline Introduction: Setting the stage HDTV HRCs Test methodologies Key challenges Proposed VQEG approach Final thoughts * Final thoughts HD TV metrics are in urgent need. We should test existing models for their correlations before seeking new ones. HD testing is happening now in organizations. A timely standardized metric will have the greatest impact. VQEG should complete testing the existing models in the next 6 months. * Next steps Off line meetings to coordinate HD testing activities. Date for next meeting? Action items captured so far: * Calibration of model inputs Peak video level +/- 10% Block level +/- 10% Horizontal shift +/- 1 pixel Vertical shift +/- 1 line Horizontal cropping 30 pixels Vertical cropping 20 lines No horizontal or vertical re-scaling Temporal alignment +/- 1 frame excluding dropped/repeated