Performance Validation and Analysis for Multi-Method Fusion Based Image Quality Metrics in A New Image Database

摘要

Considering that there is no single full reference image quality assessment meth-od that could give the best performance in all situations, some multi-method fusion metrics were proposed. Machine learning techniques are often involved in such multi-method fusion metrics so that its output would be more con-sistent with human visual perceptions. On the other hand, the robustness and generalization ability of these multi-method fusion metrics are questioned because of the scarce of images with mean opinion scores. In order to compre-hensively validate whether or not the gener-alization ability of such multi-method fusion IQA metrics are satisfying, we construct a new image database which contains up to 60 refer-ence images. The newly built image database is then used to test the generalization ability of different multi-method fusion IQA metrics. Cross database validation experiment indi-cates that in our new image database, the per-formances of all the multi-method fusion IQA metrics have no statistical significant different with some single-method IQA metrics such as FSIM and MAD. In the end, a thorough anal-ysis is given to explain why the performance of multi-method fusion IQA framework drop significantly in cross database validation.