基于大脑层状皮质模型的立体图像质量评价

发布时间：2018-02-19 18:57

  本文关键词： 质量评价 人脑 层状皮质模型 视觉感知　出处：《光电子·激光》2017年05期 　论文类型：期刊论文

【摘要】：通过模拟人脑视觉神经接收视觉信息形成表面感知的处理机制,提出一种基于大脑层状皮质模型的全参考立体图像的图像质量评价(IQA)方法。首先,分析大脑形成表面感知的过程,提出可运用于立体图像的IQA的层状皮质模型;然后依据模型得到各层的响应输出,构建感知特征向量;最后利用机器学习算法,建立特征和质量的关系模型,预测立体图像质量。实验结果表明,本文方法在对称立体图像库上的Pearson线性相关系数(PLCC)和Spearman等级系数(SROCC)高于0.91,在非对称库上高于0.93。与现有的相关方法相比,本文方法与主观评价更加吻合,更适合立体图像的评价和优化。
[Abstract]:By simulating the processing mechanism of human visual nerve receiving visual information to form surface perception, an image quality evaluation method based on the layered cortex model of brain is proposed. Firstly, IQA is used to evaluate the image quality of all reference stereoscopic images. By analyzing the process of the formation of surface perception in the brain, a layered cortical model of IQA can be applied to stereoscopic images is proposed. Then, the response output of each layer is obtained according to the model, and the perception feature vector is constructed. Finally, the machine learning algorithm is used. The relationship model of feature and quality is established to predict the quality of stereo image. The experimental results show that, In this paper, the Pearson linear correlation coefficient and Spearman rank coefficient are higher than 0.91in symmetric stereo image database and 0.93in asymmetric database. Compared with the existing correlation methods, the method in this paper is more consistent with subjective evaluation. More suitable for stereo image evaluation and optimization.
【作者单位】： 宁波大学信息科学与工程学院;
【基金】：国家自然科学基金(61271021)资助项目
【分类号】：R741.04;TP181
，

本文编号：1517869