局部模糊检测优化算法

目前基于特征的局部模糊检测算法为了优化特征响应需要在多尺度下重复计算局部模糊特征,且邻接关系复杂,导致计算量大,时间效率低.针对上述时间问题,本文提出一种利用单层垂直上下文的局部模糊检测优化算法.首先提取图像块重尾分布、峰度、功率谱、线性滤波等模糊特征,然后使用贝叶斯法学习模型,计算后验概率作为初步估计模糊响应,最后本文提出将邻近像素点的模糊响应信息作为上下文更新像素点自身响应信息,增加上下文支撑域的尺寸以更充分的考虑周围信息,使用一个相互垂直的一维上下文以减小计算量,从而构造新的能量函数进行全局优化,通过最小化能量函数得到最终的模糊响应.实验表明,本文算法能有效检测图像的局部模糊,并提高检测的时间效率.

The Viewpoint of Categorization Theory on the Nature of Reading Comprehension

This paper is intended to reveal the likelihood that conceptual categorization can be used to understand a text by reconstructing the semantic categories through which the author＇s meaning is conveyed, and proposes an alternative way to look into reading comprehension. It is proposed that categorization can be taken as an alternative approach to second/foreign language reading instruction. That is, while reading comprehension is defined in terms of the ability to recognize the inclusion and membership properties of contextually determined semantic categories in a text, the learner needs to arrange the events, actions, or concepts into a structured unit, both horizontally and vertically. Categorization theory will be introduced in relation to Rosch famous studies （1973, 1975）, examples taken from a graded reader will be illustrated as how to identify items with category structure, and finally issues that are not addressed in this paper will be discussed.