基于残差图像的迭代终止条件及其在NCSR算法中的应用

现有图像去模糊(deblurring)算法多依赖于正则化技术以迭代逼近求解最优目标函数方式实现关于原图像的最佳估计,由于缺乏迭代终止判定条件(ITC),这些算法通常采用固定迭代次数实现以至执行效率不高,同时所获得图像的质量在很多时候也未必是最佳的。鉴于在迭代过程中每一步所获得的中间估计图像经与模糊核卷积后和模糊图像之间残差图像的亮度值具有显著的高斯分布特点,提出使用广义高斯分布(GGD)模型为迭代过程中的残差图像建模并以GGD模型参数值作为衡量去模糊效果的度量(DM,deblurring measure)。基于DM,在保障去模糊图像质量的前提下,设计了ITC自适应地终止迭代过程以提高去模糊算法的计算效率。在经典的非局部集中稀疏表示(NCSR,nonlocally centralized sparse representation)去模糊算法上完成的大量的实验表明,对于运动、高斯和失焦3种典型的模糊失真,新提出的ITC能够有效判定在每一步迭代过程中所获得的估计图像是否已达到最佳的图像质量,从而实现在保障去模糊图像质量的前提下大幅度提高NCSR算法计算效率的目的。所提出的ITC判定方法具有普适性,调整相关参数后也可以应用于其它迭代型的去模糊算法。     

模糊综合评判在葡萄酒品尝中的应用

葡萄酒感官品尝是鉴定葡萄酒质量优劣的主要方法，但综存在一定的主观性和不确定性，针对这一点，本文运用模糊数学中多因素综合评判的方法评价了几种样品，结果表明综合评判和感官品尝有着一致的趋势且综合评判能更好地反映品偿鉴定的离散程度。     

The effects of metal impurities in poly[(2,5-bis(3-decylthiophen-2-yl)thieno[2,3-b]thiophene] on field-effect transistor properties

We have used particle induced X-ray emission analysis and particle induced gamma-ray emission analysis to determine the elemental impurity concentrations in poly[(2,5-bis(3-decylthiophen-2-yl)thieno[2,3-b]thiophene] samples that have undergone different washing and extraction procedures to remove impurities. Field-effect transistors (FETs) were fabricated from the materials and their electrical characteristics show no significant differences between the devices made from different material samples. Reducing the metal residue levels below the ones measured in the starting material (300 mg/kg Fe, 7 mg/kg Zn, 3000 mg/kg Pd and 12000 mg/kg Sn) does not improve the FET performance. This suggests that it is not necessary to completely remove metal residues in the polymer for FET applications.     

黄池沟配水枢纽分水池侧槽退水道水力特性试验研究

结合黄池沟配水枢纽对分水池侧槽退水道的设计要求,通过水工模型试验对侧槽退水道的水力特性进行了研究,测量了退水道水流流速、压强、水深等水力学参数,对侧堰过流能力、水面线、陡坡段水流空化及消力池流态进行了分析评价。结果表明：设计流量下侧堰满足过流能力要求,边墙高度满足过流水深要求;秦岭隧洞不会出现明满流交替现象;陡坡段受空化水流破坏的可能性较小;消力池流态差,大流量时不满足水跃消能要求,建议通过进一步的模型试验来重新设计消力池。     

Cost-effectiveness of evaluating the new technologies

The problem of evaluating cost-effectiveness claims is complex and not readily solved. However, such evaluation represents an important direction for technology assessment as resources become more scarce. Neumann et al.'s analysis (1996) represents one solution to the relatively simple problem of pharmacoeconomic studies. In addition, Fryback and Thornbury included a useful method for approaching the cost-effectiveness of all medical technologies. I note that the cost-effectiveness cutoff of $50,000 per quality-adjusted life year is an incremental one, meaning that new technology must be substantially more cost-effective than older technology. This is currently not the case where incremental improvements over previous technology are often quite small. Interested readers are referred to standard tests in the field, including Drummond, Stoddart, and Torrance (1987), Eisenberg (1986), and Sox, Blatt, Higgins, and Marton (1988).     

The relation between feelings of knowing and the number of neighboring concepts linked to the test cue

We investigated whether feeling-of-knowing judgments are influenced by the number of different neighboring concepts linked to the test cue in long-term memory as measured using association norms. The purpose was to evaluate contrasting predictions made by the partial-retrieval hypothesis and the competition hypothesis. The partial-retrieval hypothesis assumes the more neighboring concepts activated by the test cue, the higher the feeling of knowing. In contrast, the competition hypothesis assumes that feelings of knowing are sensitive to competition between neighboring concepts, and it predicts that the fewer neighboring concepts activated by the cue, the higher the feeling of knowing. The findings were compatible with the competition hypothesis showing that both feeling-of-knowing and prediction-of-knowing ratings always were higher, the fewer different concepts were linked to the test cue. We obtained an identical pattern of results using different kinds of cues including taxonomic category names, ending sounds, and meaningfully related associates. We consider different ways that these findings could be reconciled with the partial-retrieval hypothesis, and we also discuss implications for other explanations of feeling-of-knowing effects.