%0 Journal Article
%T 计算视觉核心问题：自然图像先验建模研究综述
%A 孙必慎
%A 石武祯
%A 姜峰
%J 智能系统学报
%D 2019
%R 10.11992/tis.201804019
%X 视觉先验是计算机视觉的核心问题之一，是认知心理层面、系统神经层面与计算视觉层面研究的交合点，涉及各个层面研究的理解与综合。视觉先验功能模拟方面，以自然图像信息为对象，挖掘自然图像一般性规律并将其数学形式化为可计算的图像模型，为众多图像处理与计算机视觉智能应用提供算法和支撑。本文对自然图像先验建模研究各流派工作进行了全面的剖析，并展示了自然图像先验建模工作在视觉信息增强和编码等方向的前瞻性应用。</br>One of the core problems in computer vision is that the visual prior is the point of intersection of the cognitive psychological level, systematic neural level, and computer vision level, and requires an understanding and synthesis of the three. Simulations of the visual prior function are performed to explore and formalize the general rules for natural images that support various applications in image processing and computer science. In this paper, we comprehensively analyze the work of various schools of natural image priori modeling and discuss the prospective application of natural image prior modeling in visual information enhancement and coding
%K 计算机视觉
%K 图像先验
%K 稀疏表示
%K 局部平滑
%K 非局部自相似
%K 压缩感知
%K 深度学习
%K 卷积神经网络&lt
%K /br&gt
%K computer vision
%K image prior
%K sparse representation
%K local smoothness
%K non-local self-similarity
%K compressed sensing
%K deep learning
%K convolutional neural network
%U http://tis.hrbeu.edu.cn/oa/darticle.aspx?type=view&id=201804019