留学生硕士论文 英国论文 日语论文 澳洲论文 Turnitin剽窃检测 英语论文发表 留学中国 欧美文学特区 论文寄售中心 论文翻译中心

Bussiness ManagementMBAstrategyHuman ResourceMarketingHospitalityE-commerceInternational Tradingproject managementmedia managementLogisticsFinanceAccountingadvertisingLawBusiness LawEducationEconomicsBusiness Reportbusiness planresearch proposal


ResumeRecommendation LetterMotivation LetterPSapplication letterMBA essayBusiness Letteradmission letter Offer letter



英语论文开题报告英语毕业论文写作指导英语论文写作笔记handbook英语论文提纲英语论文参考文献英语论文文献综述Research Proposal代写留学论文代写留学作业代写Essay论文英语摘要英语论文任务书英语论文格式专业名词turnitin抄袭检查








高效的可视化搜索对象的影片:Efficient Visual Search for Objects in Videos

论文作者:留学生论文论文属性:学期论文 termpaper登出时间:2011-02-15编辑:anterran点击率:11582

论文字数:5142论文编号:org201102150926302217语种:英语论文 English地区:英国价格:免费论文

关键词:Object recognitiontext retrievalviewpointscale invariance

Efficient Visual Search for Objects in Videos
Visual search using text-retrieval methods can rapidly and accurately
代写留学生论文locate objects in videos despite changes in camera viewpoint,
lighting, and partial occlusions.
By Josef Sivic and Andrew Zisserman
ABSTRACT | We describe an approach to generalize theconcept of text-based search to nontextual information. Inparticular, we elaborate on the possibilities of retrievingobjects or scenes in a movie with the ease, speed, and accuracywith which Google [9] retrieves web pages containing particularwords, by specifying the query as an image of the object orscene. In our approach, each frame of the video is represented
by a set of viewpoint invariant region descriptors. Thesedescriptorsenable recognition to proceed successfully despitechanges in viewpoint, illumination, and partial occlusion.Vector quantizing these region descriptors provides a visualanalogy of a word, which we term a Bvisual word.[ Efficientretrieval is then achieved by employing methods from statisticaltext retrieval, including inverted file systems, and text anddocument frequency weightings. The final ranking also dependson the spatial layout of the regions. Object retrievalresults are reported on the full length feature films BGroundhog
Day,[ BCharade,[ and BPretty Woman,[ including searches from
within the movie and also searches specified by external
images downloaded from the Internet. We discuss three
research directions for the presented video retrieval approach
and review some recent work addressing them: 1) building
visual vocabularies for very large-scale retrieval; 2) retrieval of
3-D objects; and 3) more thorough verification and rankingusing the spatial structure of objects.

KEYWORDS | Object recognition; text retrieval; viewpoint and scale invariance

The aim of this research is to retrieve those key frames andshots of a video containing a particular object with theease, speed, and accuracy with which web search enginessuch as Google [9] retrieve text documents (web pages)containing particular words. An example visual object
query and retrieved results are shown in Fig. 1. This paper
investigates whether a text retrieval approach can besuccessfullyemployed for this task.
Identifying an (identical) object in a database of imagesis a challenging problem because the object can have adifferent size and pose in the target and query images, andalso the target image may contain other objects (Bclutter[)that can partially occlude the object of interest. However,successful methods now exist which can match an object’svisual appearance despite differences in viewpoint, lighting,
and partial occlusion [22]–[24], [27], [32], [38], [39],
[41], [49], [50]. Typically, an object is represented by a set
of overlapping regions each represented by a vector
computed from the region’s appearance. The region
extraction and descriptors are built with a controlled
degree of invariance to viewpoint and illumination
conditions. Similar descriptors are computed for all images
in the database. Recognition of a particular object proceeds
by nearest neighbor matching of the descriptor vectors,
followed by disambiguating or voting using the spatial
consistency of the matched regions, for example by
computing an affine transformation between the query
and target image [19], [22]. The resul论文英语论文网提供整理,提供论文代写英语论文代写代写论文代写英语论文代写留学生论文代写英文论文留学生论文代写相关核心关键词搜索。

共 1/19 页首页上一页1234567下一页尾页


    英国英国 澳大利亚澳大利亚 美国美国 加拿大加拿大 新西兰新西兰 新加坡新加坡 香港香港 日本日本 韩国韩国 法国法国 德国德国 爱尔兰爱尔兰 瑞士瑞士 荷兰荷兰 俄罗斯俄罗斯 西班牙西班牙 马来西亚马来西亚 南非南非

       Europe (24-hours)
       china (24-hours)