Subjective-evaluation method for quality of coded images

Subjective-evaluation tests for the quality of coded images are conducted usually using the double-stimulus continuous quality-scale method recommended by the ITU-R. A new method named "the modified EBU method", which is different from the EBU method in the presentation of the test sequences, is presented in this paper. The modified EBU method's ability to discriminate small quality differences is examined by comparison with the relative scale of the double-stimulus continuous quality-scale method, for the significant level of 0.05, using four sequences. It is seen for 15 experts that significant differences are not detected though test sequences are apparently degraded in one condition for each of two sequences with the double-stimulus continuous quality-scale method, while a significant difference is not detected in only one condition for one sequence with the modified EBU method. On the other hand, it is seen for 20 non-experts that significant differences are not detected in a total of five conditions for three sequences with the double-stimulus continuous quality-scale method, while significant differences are not detected in one condition for each of two sequences with the EBU method. Therefore, it is clarified that the modified EBU method are more suitable for small impairments than the relative scale of the double-stimulus continuous quality-scale method. >