Learning in syntactic recognition of symbols drawn on a graphic tablet

This paper deals with on-line recognition of handprinted characters and other common symbols, drawn on a graphic tablet. The system relies on meaningful, morphological features and a syntactic, or structural, approach. The first step, after preprocessing, is the representation of a character by the time sequence of its features. This sequence is extracted by means of a finite automation. The representation is completed, when necessary, by topological relations between features. The system can learn to recognize new symbols during normal use, since it needs very few samples for training.     

Understanding small graphical symbols: a cross-cultural study

Graphical symbols such as icons or pictograms as interfaces in modern technological devices have become quite common. Although generally designed in western countries, their use can be found worldwide from audio-visual appliances to computers and computer-related devices. A basic ergonomic principle is to involve the potential users in the design and evaluation stages especially when the user groups are quite diverse. This study was part of an international project on evaluating telecommunication symbols. One of the major objectives was to test different graphical symbols (of the videophone), designed, and tested in Western Europe using subject groups from Asia, Europe, and the United States. This paper would discuss the major portions of the tests involving US and Swedish subjects. Performance indices used were hits, certainties, confusions, and semantic differential ratings. They were useful in analysing how the symbols were recognised, confused, and perceived by different subject groups. They also helped detect differences between groups which otherwise seemed to have similar test results. The results showed differences in patterns of ratings, which may be culturally linked and could help determine aspects of symbol design and usage that may be more helpful in designing instructions, learning aids, etc. Awareness of such subject bias and their implications are important on how one interprets the test results.

Relevance to industry

The study outcome can be very helpful for industries in designing and evaluating candidate graphical symbols across various types of consumer products for worldwide use.     

A graphic model of the human hand using CATIA

Anthropometric measurements were made of the hands of 32 subjects. Data from this survey were used in the design of a graphic model of the male human hand. The model was created in the computer-aided design (CAD) package CATIA and reproduced any desired overall hand size. It was made up of 24 solid segments with 23 degrees of freedom at 17 moveable joints. Spans and grasping sizes of the model hand were examined. The representation of the model appeared realistic in several postures, but the model was found to allow some motions which would be impossible for an actual hand. The model could be used to help a designer visualize hand-held products and controls in use.     

A framework for modelling graphic interactions

In the present-day graphics field, a proper methodology for designing input constructs of a general purpose graphics system is needed. In this direction, this paper proposes a framework where the graphics interaction programming is formulated as event-driven procedures. The end-user actions on the input tools are represented as events and the programreactions to these events as action-processes. The constructs for programming in this framework are implemented as a part of a graphics package SODDI. The resultant programming facilities are compared with the other existing proposals, based on the conceptual differences of the respective frameworks as well as on the experiences in using the constructs. The data-structures used for implementing these constructs are also presented.     

一种基于特征符号的网页主题信息抽取方法

随着Internet网络的日益普及,Web上的海量数据给文本挖掘尤其是网页主题提取带来了更多的挑战,现有的文本提取方法在保证高准确率的同时无法满足Web挖掘方法的通用性。通过对Web网页结构进行研究,对网页生成树模型进行了改进,找到网页结构的通用规则,提出一种基于特征符号的提取方法CECS（content extraction characteristic symbols）,结合相关度对网页主题内容进行提取。实验证明,所提算法具有很高的准确性和通用性。     

基于相关系数符号的快速匹配算法*

针对经典灰度相关匹配算法耗时多的缺点,提出一种减少计算次数、提高匹配效率的新方法.该方法是先构造一幅相关图像,然后计算它与基准子图像的互相关数据并利用其符号分离候选匹配点,最后在匹配过程中,根据实时图像与相关图像的相关系数符号来选定匹配点区域,从而减少匹配计算次数,提高匹配速度.理论分析和实验表明,该方法不但能够保持经典灰度相关匹配算法具有较高的匹配精度和较强的抗噪能力的特点,而且大大缩短了匹配时间,使得经典灰度相关算法能够在实际中得以应用.     

Experimental studies of control problem-solving processes using a graphic mnemonic display

Experiments were conducted to examine different cognitive strategies in control problem-solving using a graphic mnemonic display typical for the control rooms of complex technological systems in power, chemical, and metallurgical industries. The law of plurality of human work strategies (Vehda 1985) was used as a fundamental assumption for this study. Detailed studies of the cognitive strategies' characteristics are especially important for design of the information display, and further development of the theory of transformation dynamics (Venda and Venda 1991,1992). During experiments at a Moscow power plant three different strategies were identified (S_sy,S_w and S_p). Detailed follow-up experiments were conducted in the laboratory using a mock-up of a section of the power plant mnemonic display. Subject performance was analysed using eye movement registration, EEG recording, self-reporting, questionnaires, larynx myography, and time and error analysis. The follow-up experiments found three distinct strategies (S_ι, S_y and S_w). Laboratory assessment of information displays was validated only for the strategies S_y and S_w which were common for the real operators and subjects.     

基于轮廓特征的印刷公式粘连符号的切分

在印刷体数学公式识别中,不能准确地切分粘连符号是造成识别错误的主要原因之一。针对这种情况,提出了一种基于轮廓特征切分粘连符号的方法。根据轮廓特征及宽高比形成切分路径,然后对粘连字符切分。实验表明,这种方法使识别率得到了明显提高。     

基于图表法的粗糙集约简

在文[1]由差别矩阵计算信息系统核的基础上得到未定差别项组,提出了基于表格因子图的图解法。利用该方法能准确地求出信息系统中所有的最小子集,且计算量少于由定义来约简。举例说明了该方法的有效性。     

乐谱图像乐符分割技术*

提出基于图段拓扑关系的谱线删除方法,以避免谱线过删除现象;提出双向游程编码结合使用的符干分割方法,克服了现有方法对复杂音符适应性差、分割结果不完整等缺陷;提出音符先验知识引导下的符头切割与检测算法,以解决粘连符头的切分问题;提出基于块状体分割和特征检测的符梁分割算法,设计了适用于乐谱版面的文字和线条提取算法。该方法应用在乐谱识别系统中分割乐符具有良好的性能,尤其对乐谱内容复杂、乐符排列密集等情况有较强适应能力。     

Hong Kong Chinese and Korean comprehension of American security safety symbols

This research evaluated Hong Kong Chinese and Korean comprehension of American security safety symbols, and how successfully they could guess the meaning of the symbols in relation to their ratings of the appropriateness of the design of each of the symbols. Symbol comprehension scores, symbol guessing scores, design appropriateness ratings, and demographic information were obtained for 81 Hong Kong Chinese and 60 Koreans. For all the symbols tested, comprehension scores for the Hong Kong Chinese and Koreans were much lower than for Americans. The finding that Americans were better at interpreting American symbols than Hong Kong Chinese and Koreans indicates that problems are likely to arise if such symbols are used by non-Americans. Gender did not affect guessing performance, and success at guessing symbols was not related to the subjective ratings of design appropriateness for the symbols. The findings here reveal the importance of developing security safety symbols with the end users in mind.

Relevance to industry

Designing effective pictorial security safety symbols to cover the many potential situations and scenarios for a specific population is a difficult business. The findings of this study underline the importance of developing security safety symbols with the end users in mind. The results provide useful information to assist in the design of more user-friendly security safety symbols.     

A pre-distortion based design method for digital audio graphic equalizer

The traditional design method for digital audio graphic equalizer using infinite impulse response filters usually has some deficiencies, including center frequency shift, narrower bandwidth at high frequency and inaccurate gain. In this paper, an improved method based on the modified bilinear transformation is proposed to design a digital audio graphic equalizer. The new bilinear mapping can compensate the center frequency shift, and pre-distorting the quality factors and optimizing the gains can correct the bandwidth and gain of each sub-band respectively. Experimental results reveal that both center frequency and bandwidth of the proposed digital graphic equalizer are strictly equal to the desired ones, and the average gain error decreases at least 2 dB.     

基于GPU的串匹配算法研究

BF算法是串匹配算法中最基础的算法，但它是串行算法，不适合图形处理器（Graphic Processing Unit, GPU）的体系结构。结合GPU的特殊体系结构，通过数据存取方式和计算策略的改进，充分利用了GPU的并行处理能力，从而基于GPU实现了BF算法。实验结果表明基于GPU的并行算法能够取得较好的加速比，同时也给出了在现有GPU架构上有效实现通用计算的瓶颈。     

User trials on the colours and symbols for electronic charts

The Electronic Chart Display and Information System (ECDIS) contributes to safe navigation by enabling the mariner to follow a planned route precisely, using positioning and radar information integrated on an electronic chart. The need to ensure that ECDIS is safe to use has prompted the development of performance standards for the presentation of information on electronic charts. As a step in working towards these presentation standards, user trials were conducted on shore and at sea using an ECDIS display to investigate presentation issues, which included colour combinations and symbols for ships and tracks, chart features, radar overlays, scale indicators and also display brightness for day and night viewing. Results of questionnaires indicated mariners' preferences and rated visibility of various colours and symbols. As a result of the shore trials, improvements were made to some of the symbols, which were then reevaluated during sea trials. Changes to the colours and symbols resulting from these trials will be included in the International Hydrographic Organization's Provisional Colour and Symbol Standards for ECDIS.     

诱导系统数据模型在城市指路标志智能布设中的应用

为了实现指路标志系统的智能化管理, 以指路标志诱导系统数据模型为基础, 提出了基于道路指示等级的指路标志自动布设算法。该算法是以道路交叉口为研究对象, 根据指路标志指引信息与路网拓扑的关系, 将指路标志指引信息分成连接性标志和方向性标志两种, 定义了基于道路指示等级的方向性标志选择规则, 并设计了生成方向性标志的算法模型。最后以广州大学城为实验区域, 实验结果表明诱导系统数据模型可有效支持指路标志的生成及分析评价。     

三维静态军标的实时生成与标绘

三维静态军标的生成与标绘是军事信息可视化系统的重要内容,生成与标绘方法的优劣直接关系到可视化系统的运行效率和执行效果.在现有方法的基础上提出了一种结合实体模型和公告板技术的实时生成与标绘三维静态军标的方法.难点在于将GDI实时生成的二维军标图像经过提出的一种图像算法的处理产生公告板待显示的具有三维效果的军标图像.实验结果表明该方法的三维军标标绘速度快,占用内存空间小,在应用场景中三维军标的战术意义明确,真实感、立体感强,很好地满足了实际可视化系统的需求.     

照度场中M阵列符号图像阈值分割方法研究

针对不均匀性照度场的干扰问题,从M阵列符号源图像阈值分割出发,摒弃前期研究中惯用的固定阈值分割方法,提出了以某一点为中心的邻域亮度和阈值对应的分布关系构建成该符号源图像的阈值分布场的自适应阈值分割方法,由此能得出在该照度场中的符号图像的合理阈值.实验结果表明,新的阈值分割方法可进一步提高符号的识别效率,并为后期的三维拼接工作奠定了良好的基础.     

基于GPU的光子映射并行化算法

针对串行情况下光子映射算法速度慢的问题,对光子映射算法并行化进行可行性分析,充分利用图像处理器(GPU)的统一设备计算架构(CUDA)的并行和计算能力,实现光子映射算法的并行化。同时针对算法中光子发射追踪阶段生成GPU线程数与光子数相同的方法的不足以及平均分配方法所造成的资源浪费等,提出线程之间协同工作的方法并采用动态平衡处理,使光子渲染速度提升了将近一倍。实验结果证明了多线程间协同工作及动态平衡相结合方法的有效性。