非平稳确定性信号与非平稳随机信号统一分类法的探讨

根据信号通过线性系统可产生新信号的理论,对非平稳确定性信号与非平稳随机信号的统一分类方法进行了深入研究,分别给出格林函数描述的线性时变系统法、调制函数作用于边界稳定线性系统的自激振荡正弦波法、随机输入线性时不变系统与边界稳定的与不稳定的线性系统组合法。这3种分类法便于对非平稳确定性信号与非平稳随机信号进行统一分类研究,并具有比较全面与系统性好的优点。     

基于块分类稀疏表示的图像融合算法

针对基于单一字典训练稀疏表示的图像融合算法忽略图像局部特征的问题,提出了基于块分类稀疏表示的图像融合算法。此算法是根据图像局部特征的差异将图像块分为平滑、边缘和纹理三种结构类型,对边缘和纹理结构分别训练出各自的冗余字典。平滑结构利用算术平均法进行融合,边缘和纹理结构由对应字典利用稀疏表示算法进行融合,并对边缘结构稀疏表示中的残余量进行小波变换融合。实验结果证明,该算法相对于单一字典稀疏表示算法,在融合图像的主观评价和客观评价指标上都有显著改进,并且算法速度也有提高。     

Stakeholder Representations of Gentrification in Amsterdam and Berlin: A Marginal Process?

In recent years, several studies have highlighted how gentrification strategies are imposed under the discursive umbrella of ‘social mixing’. However, most evidence is based on Anglo-Saxon experiences. This paper sets out to expand the geography of gentrification by looking at the representation of processes and policies of gentrification as put forward by key stakeholders in Nord-Neukölln (Berlin) and Indische Buurt (Amsterdam). It shows that in both contexts, stakeholders and policy documents engage with the concept of gentrification, rather than avoid it. Due to public-policy influence and local criticisms, this engagement differs between both cases. In Nord-Neukölln, the term is heavily contested and policy-makers attempt to refute accusations of gentrification, while in the Indische Buurt, the process is explicitly pursued as a positive policy instrument by policy-makers. Different representations within each case are shown to be influenced by the characteristics of in-moving and out-moving residents; the employed timeframe and the perceived influence of institutions on urban regeneration.     

基于稀疏表达残差的自然场景运动目标检测

本文提出一种基于稀疏表达残差的非参数化运动目标检测算法,在假设前景变化相对静态背景可以视为稀疏残差的基础上,采用视频前n帧初始化稀疏表达字典;利用字典对后续视频帧进行重构,提取每帧的重构残差;结合基于光照强度的全局阈值矩阵,将残差图像二值化,提取图像前景;利用前景区域和边缘点关系剔除ghost区域;采用增量PCA(Principal Component Analysis)算法和保守更新的思想对背景模型进行更新.在changedetection.net提供的shadow数据集上实验表明,采用全局更新和残差计算的方法,可以有效的解决由于自然场景光线变化导致的阴影变化,并且对自然场景中背景的小幅度抖动和相机抖动等问题也具有一定的抵抗能力.     

基于Gabor特征和字典学习的高斯混合稀疏表示图像识别

为了克服图像识别中光照,姿态等变化带来的识别困难,同时提高稀疏表示图像识别的鲁棒性,本文提出了一种基于Gabor特征和字典学习的高斯混合稀疏表示图像识别算法.高斯混合稀疏表示是基于最大似然估计准则,将稀疏保真度表示为余项的最大似然函数,最终识别问题转化为求解加权范数的优化逼近问题.本文算法首先提取图像的Gabor特征;然后对Gabor特征集进行字典学习,由于在学习过程中引入了Fisher准则作为约束,学习得到具有类别标签的新字典;最后使用高斯混合稀疏表示识别方法进行分类识别.在3个公开数据库(人脸数据库AR库和FERET库以及USPS手写数字库)上的实验结果验证了该算法的有效性和鲁棒性.     

面向目标检测的稀疏表示方法研究进展

目标检测作为图像理解的一个基础而重要的课题深受国内外学者的重视,在军事和民用中具有广泛应用.应用背景的多样性和复杂性使得传统目标检测算法难以克服复杂背景、噪声干扰、光照变化以及非刚体形变、遮挡、弱特征、尺度、视角和姿态变化等因素的影响.近些年来发展起来的稀疏表示方法为图像处理及目标检测研究提供了新的思路,本文概述了稀疏表示基本概念和理论研究进展,综述了稀疏表示方法在目标特征学习、目标分类器和滤波器设计以及多源信息融合目标检测等目标检测领域中的国内外重要研究进展,并展望了稀疏表示方法在目标检测领域的发展方向.     

The role of stimulus comparison in human perceptual learning: Effects of distractor placement.

Within-subjects procedures were used to assess the influence of stimulus comparison on perceptual learning in humans. In Experiment 1, participants received intermixed (A, A′, A, A′,…) or blocked (B, B,…, B′, B′,…) exposure to pairs of similar female faces. In a subsequent same/different discrimination task, participants were more accurate when the test involved A and A′ than when it involved B and B′ (or novel faces: C and C′). This perceptual learning effect was reduced by placing a visual distractor (*: either another face or a checkerboard) between successive presentations of the faces during the exposure stage (e.g., A – * – A′). The attenuation of the intermixed versus blocked difference was particularly marked when faces were used as the distractor. In Experiment 2, this reduction in perceptual learning was more marked when * was positioned between the pairs of intermixed faces (i.e., A – * – A′) than when it preceded and succeeded those faces (i.e., * – A – A′ – *). These results provide the first direct evidence that the opportunity to compare stimuli plays a causal role in supporting perceptual learning. They also support the specific view that perceptual learning reflects an interaction between a short-term habituation process, that ordinarily biases processing away from the frequently presented common elements and toward their less frequently presented unique elements, and a long-term representational process that reflects this bias. (PsycINFO Database Record (c) 2011 APA, all rights reserved)     

Specific previous experience affects perception of harmony and meter.

Prior knowledge shapes our experiences, but which prior knowledge shapes which experiences? This question is addressed in the domain of music perception. Three experiments were used to determine whether listeners activate specific musical memories during music listening. Each experiment provided listeners with one of two musical contexts that was presented simultaneously with a melody. After a listener was familiarized with melodies embedded in contexts, the listener heard melodies in isolation and judged the fit of a final harmonic or metrical probe event. The probe event matched either the familiar (but absent) context or an unfamiliar context. For both harmonic (Experiments 1 and 3) and metrical (Experiment 2) information, exposure to context shifted listeners' preferences toward a probe matching the context that they had been familiarized with. This suggests that listeners rapidly form specific musical memories without explicit instruction, which are then activated during music listening. These data pose an interesting challenge for models of music perception which implicitly assume that the listener's knowledge base is predominantly schematic or abstract. (PsycINFO Database Record (c) 2011 APA, all rights reserved)     

Sustained effects of adaptation on the perception of familiar faces.

Figural aftereffects are commonly believed to be transient and to fade away in the course of milliseconds. We tested face aftereffects using familiar faces and found sustained effects lasting up to 1 week. In 3 experiments, participants were first exposed to distorted pictures of famous persons and then had to select the veridical face in a 2-alternative forced choice task. Veridicality aftereffects were found in a direction opposite to the adapting distortion; these effects generalized to other pictures of the same individual and also to pictures of celebrities that had not been shown during adaptation. The findings support hierarchical theories of norm-based face coding and suggest that face adaptation effects have a representational basis. They also point toward multiple timescales in the operation of adaptation mechanisms, thereby providing a link between high-level adaptation and more general aspects of neuro-cognitive plasticity, that is, learning and memory. (PsycINFO Database Record (c) 2011 APA, all rights reserved)