Similar Reference Image Quality Assessment: A New Database and A Trial with Local Feature Matching

摘要

Conventionally, the reference image for image quality assessment (IQA)rnis completely available (full-reference IQA) or unavailable (no-reference IQA).rnEven for reduced-reference IQA, the features that are used to predict image qualityrnare still extracted from the pristine reference image. However, the pristine referencernimage is always unavailable in many real scenarios. In contrast, it is convenient tornobtain a number of similar reference images via retrieval from the Internet. Thesernsimilar reference images may share similar contents and scenes with the image to bernassessed. In this paper, we attempt to discuss the image quality assessment problemrnfrom the view of similar images, i.e. similar reference IQA. Although the similarrnreference images share similar contents with the degraded image, the differencernbetween them still cannot be ignored. Therefore, we propose an IQA frameworkrnbased on local feature matching, which can help to identify the similar regions andrnstructures. Then the IQA features are computed only from these similar regions tornpredict the final image quality score. Besides, since there is no IQA databases for thernsimilar reference IQA problem, we establish a novel IQA database that consists ofrn272 images from four scenes. The experiments demonstrate that the performance ofrnour scheme goes beyond state-of-the-art no-reference IQA methods and some fullreferencernIQA algorithms.