Blind Image Quality Assessment via Deep Learning
Blind Image Quality Assessment via Deep Learning
复制标题
通过深度学习进行盲图像质量评估
DOI：
10.1109/tnnls.2014.2336852
复制
发表时间：
2015-06-01
影响因子：
10.4
通讯作者：
Li, Xuelong
中科院分区：
文献类型：
--
作者：
Hou, Weilong;Gao, Xinbo;Li, Xuelong
本文研究了如何从语言描述中学习规则来盲评价图像的视觉质量。大量的心理学证据表明，人类更喜欢进行定性评估，而不是数字。然后将定性评价转换为数值分数，以公平地基准客观图像质量评估（IQA）指标。近年来，许多基于学习的IQA模型通过分析图像到数字评分的映射关系而被提出。然而，学习的映射很难足够准确，因为在从语言描述到数字分数的这种不可逆转换中丢失了一些信息。在本文中，我们提出了一个盲IQA模型，直接学习定性评价和输出数值分数的一般利用和公平的比较。图像由自然场景统计特征表示。训练区分性深度模型以将特征分类为五个等级，对应于五个明确的心理概念，即，优秀，好，一般，差，坏。然后应用新设计的质量池将定性标签转换为分数。分类框架不仅比基于回归的模型更自然，而且对小样本问题也具有鲁棒性。在流行的数据库上进行了深入的实验，以验证该模型的有效性，效率和鲁棒性。
This paper investigates how to blindly evaluate the visual quality of an image by learning rules from linguistic descriptions. Extensive psychological evidence shows that humans prefer to conduct evaluations qualitatively rather than numerically. The qualitative evaluations are then converted into the numerical scores to fairly benchmark objective image quality assessment (IQA) metrics. Recently, lots of learning-based IQA models are proposed by analyzing the mapping from the images to numerical ratings. However, the learnt mapping can hardly be accurate enough because some information has been lost in such an irreversible conversion from the linguistic descriptions to numerical scores. In this paper, we propose a blind IQA model, which learns qualitative evaluations directly and outputs numerical scores for general utilization and fair comparison. Images are represented by natural scene statistics features. A discriminative deep model is trained to classify the features into five grades, corresponding to five explicit mental concepts, i.e., excellent, good, fair, poor, and bad. A newly designed quality pooling is then applied to convert the qualitative labels into scores. The classification framework is not only much more natural than the regression-based models, but also robust to the small sample size problem. Thorough experiments are conducted on popular databases to verify the model's effectiveness, efficiency, and robustness.