一种小波域K-Means遥感图像分类标注算法   总被引：1，自引：0，他引：1  

彭金喜  苏远歧  薛笑荣 《教育技术导刊》2019,18(9):202-206

由于合成孔径雷达图像（遥感）的相干斑噪声数据丰富,导致传统的遥感图像分割方法分割效果不佳,采用学习理论和神经网络改善图像处理性能。根据图像统计特征,采取神经网络语义提出一种高效的图像纹理特征分割方法。首先,利用K-means聚类提取遥感图像的纹理特征,然后根据遥感图像在小波域中的分布特征对其进行滤波,最后利用语义对滤波后的遥感图像纹理特征和灰度组成的矢量进行分割归类,在遥感图像分割中快速标注分类以便于视觉分析。利用区域一致性分割分类,由聚类样本特征匹配进行图像分类标注,对变化检测进行统计分析,过分割或欠分割误差聚类样本不做标注,选取最佳样本聚类k值标注分类结果。  相似文献   

车牌识别技术中字符切割新算法     

李元金  高维春  王精明 《深圳信息职业技术学院学报》2007,5(1):32-34

先对切割下来的车牌进行水平扫描,确定车牌上文字在水平上的大致范围;然后再对确定水平范围后的车牌图像进行竖直扫描确定车牌上每个字符的大致宽度;最后把它们切割下来并进行归一化处理,为下一步车牌特征提取打下了基础。实验结果的数据表明该方法对车牌上字符的切割具有很好的效果。  相似文献   

汉语自动分词中切分歧义及处理技术     

王中立 《许昌学院学报》2006,25(2):118-121

描述了汉语自动分词中切分歧义的发现和歧义字段的消除,给出了基于词典的汉语词自动切分和基于统计的词类与词性歧义消除的模型和实现方法.  相似文献   

Unconventional word segmentation in Brazilian children’s early text production     

Jane Correa  Julie E. Dockrell 《Reading and writing》2007,20(8):815-831

An important element of learning to read and write at school is the ability to define word boundaries. Defining word boundaries in text writing is not a straightforward task even for children who have mastered graphophonemic correspondences. In children’s writing, unconventional word segmentation has been observed across a range of languages and contexts with more occurrences of hyposegmentation (failure to separate two or more written words with a space) than hypersegmentation (written words are split into more than one segment). However, it is still unclear how frequent these errors are and the relationships of these written error patterns to the child’s development in oral language, spelling and reading remains relatively unexplored. To address these issues, unconventional written lexical segmentations in Brazilian Portuguese children’s text production during their first years at primary school (Year 1 to Year 3) were examined in relation to different cognitive and linguistic measures and patterns of spelling errors. Results reveal that in Portuguese the establishment of word boundaries in written text is not explained by visuospatial skills or limitations in processing resources (working memory). In contrast higher occurrences of hyposegmentation patterns were associated with lower levels of reading, vocabulary, verbal ability and morphological awareness whereas hypersegmentations were rarer and related to lower levels of reading and morphological awareness and typically only occurred in the initial stages of learning to write (Year 1). Occurrences of hyposegmentations as well as hypersegmentations were also related to spelling errors which reflected children’s poor phonological skills.  相似文献   

基于Lucene的中文分词技术改进     

刘敏娜 《咸阳师范学院学报》2012,27(2):55-57

分析现有的词表切分法、自动切分算法的不足,通过改进Lucene系统的分词功能,设计了一个基于Lucene的中文数码产品搜索引擎,重点探讨了该搜索引擎的中文分词功能,实验证明本方法对中文词语可以设定正向匹配的字数,从而灵活的有效地进行中文分词。  相似文献   

数字图像分割方法研究     

杨建菊 《凯里学院学报》2012,30(3):119-121

图像分割即指将图像分成各具特性的区域并提取出感兴趣目标的技术,它是从图像处理到图像分析关键步骤.在对不同的图像分割方法进行分析基础上,对自适应分水岭图像分割方法进行研究.  相似文献   

结合扩展LoB滤波的距离保持水平集方法     

秦进  付茂臣  任泽民 《遵义师范学院学报》2012,14(6):85-88

二项分布拉普拉斯边缘检测算子LoB主要利用图像的二阶微分。作者在此基础上结合图像的张量信息提出了改进的LoB滤波算子,作用于距离保持水平集方法的面积项,从而构造出新的模型。与原模型相比,该模型提高了图像的演化速度,能够分割多目标物体且对具有丰富尖角细节、深度凹陷和狭窄缝隙的图像拥有较好的分割效果。  相似文献   

不同专业方向的设计基础教学研究探析     

邱勇 《四川教育学院学报》2014,(7):13-17

设计专业的基础课程设置目的在于加强学生空间造型能力方面的培养及审美意识的提高,就目前艺术设计专业普遍在设计基础课程教学中存在的单一化现象,提出应根据现有设置设计专业的具体培养方向,基于空间分类的研究来实现能力培养的细分化。并以空间形态中构成观的培养为研究重点,以空间思维转换、空间形态认知为切入点,探讨了教学体系中相对应的调整和革新。  相似文献   

基于SVM的异常行为检测在电子考场中的应用研究     

熊思 《湖北第二师范学院学报》2013,(8):42-46

电子监考系统在我国已经得到广泛应用,本文针对考场视频分析的应用需求,以智能视频监控理论为依据,研究基于视频分析的考场异常行为有效识别方法。本文基于目标对象的皮肤特征模型,设计和生成行为模型,结合SVM支持向量机,实现在视频帧序列中检测考场异常行为。该模型可以应用于考场视频监控系统中,有助于提高系统的智能化,增强考场监控能力。  相似文献   

Comparison of 3D reconstructive technologies used for morphometric research and the translation of knowledge using a decision matrix     

Charys M. Martin  Victoria A. Roach  Ngan Nguyen  Charles L. Rice  Timothy D. Wilson 《Anatomical sciences education》2013,6(6):393-403

The use of three‐dimensional (3D) models for education, pre‐operative assessment, presurgical planning, and measurement have become more prevalent. With the increase in prevalence of 3D models there has also been an increase in 3D reconstructive software programs that are used to create these models. These software programs differ in reconstruction concepts, operating system requirements, user features, cost, and no one program has emerged as the standard. The purpose of this study was to conduct a systematic comparison of three widely available 3D reconstructive software programs, Amira^®, OsiriX, and Mimics^®, with respect to the software's ability to be used in two broad themes: morphometric research and education to translate morphological knowledge. Cost, system requirements, and inherent features of each program were compared. A novel concept selection tool, a decision matrix, was used to objectify comparisons of usability of the interface, quality of the output, and efficiency of the tools. Findings indicate that Mimics was the best‐suited program for construction of 3D anatomical models and morphometric analysis, but for creating a learning tool the results were less clear. OsiriX was very user‐friendly; however, it had limited capabilities. Conversely, although Amira had endless potential and could create complex dynamic videos, it had a challenging interface. These results provide a resource for morphometric researchers and educators to assist the selection of appropriate reconstruction programs when starting a new 3D modeling project. Anat Sci Educ 6: 393–403. © 2013 American Association of Anatomists.  相似文献