Hint: We tend to remember pictures of people much better than wide open spaces CAMBRIDGE, Mass. -- Next time you go on vacation, you may want to think twice before shooting hundreds of photos of that scenic mountain or lake. A new study from MIT neuroscientists shows that the most memorable photos are those that contain people, followed by static indoor scenes and human-scale objects. Landscapes? They may be beautiful, but they are, in most cases, utterly forgettable. "Pleasantness and memorability are not the same," says MIT graduate student Phillip Isola, one of the lead authors of the paper, which will be presented at the IEEE Conference on Computer Vision and Pattern Recognition, taking place June 20-25 in Colorado Springs. The new paper is the first to model what makes an image memorable — a trait long thought to be impenetrable to scientific study, because visual memory can be so subjective. "People did not think it was possible to find anything consistent," says Aude Oliva, associate professor of cognitive science and a senior author of the paper. However, the MIT team, which also included Antonio Torralba, the Esther and Harold E. Edgerton Associate Professor of Electrical Engineering and Computer Science, and one of his graduate students, Jianxiong Xiao, was surprised to see remarkable consistency among hundreds of people who participated in the memory experiments. Using their findings from humans, the researchers developed a computer algorithm that can rank images based on memorability. Such an algorithm could be useful to graphic designers, photo editors, or anyone trying to decide which of their vacation photos to post on Facebook, Oliva says. Why we remember Oliva's previous research has shown that the human brain can remember thousands of images, with a surprising level of detail. However, not all images are equally memorable. For the new study, the researchers built a collection of about 10,000 images of all kinds — interior-design photos, nature scenes, streetscapes and others. Human subjects in the study (who participated through Amazon's Mechanical Turk program, which farms tasks out to people sitting at their own computers) were shown a series of images, some of which were repeated. Their task was to indicate, by pressing a key on their keyboard, when an image appeared that they had already seen. Each image's memorability rating was determined by how many participants correctly remembered seeing it. In general, different research subjects tended to produce similar memorability ratings. "There are always differences between observers, but on average, there is very high consistency," says Oliva, who is also a principal investigator in the computer vision group at MIT's Computer Science and Artificial Intelligence Laboratory. After gathering their data, the researchers made "memorability maps" of each image by asking people to label all the objects in the images. A computer model can then analyze those maps to determine which objects make an image memorable. In general, images with people in them are the most memorable, followed by images of human-scale space — such as the produce aisle of a grocery store — and close-ups of objects. Least memorable are natural landscapes, although those can be memorable if they feature an unexpected element, such as shrubbery trimmed into an unusual shape. Alexei Efros, associate professor of computer science at Carnegie Mellon University, says the study offers a novel way to characterize images. "There has been a lot of work in trying to understand what makes an image interesting, or appealing, or what makes people like a particular image. But all of those questions are really hard to answer," says Efros, who was not involved in this research. "What did was basically approach the problem from a very scientific point of view and say that one thing we can measure is memorability." Predicting memorability The researchers then used machine-learning techniques (a type of statistical analysis that allows computers to identify patterns in data) to create a computational model that analyzed the images and their memorability as rated by humans. For each image, the computational model analyzed various statistics — such as color, or the distribution of edges — and correlated them with the image's memorability. That allowed the researchers to generate an algorithm that can predict memorability of images the computational model hasn't "seen" before. Such an algorithm could be used by book publishers to evaluate cover art, or news editors looking for the most memorable photograph to feature on their website. Oliva believes the algorithm might also be of interest to camera manufacturers, and Isola is thinking about designing an iPhone app that could immediately tell users how memorable the photo they just took will be. For that application, the main challenge is getting the algorithm to work fast enough, Isola says. Other possible applications are clinical memory tests that more precisely reveal what aspects of visual memory are deficient in specific psychological or brain disorders, and games to help train the memory. The researchers are now doing a follow-up study to test longer-term memorability of images. They are also working on adding more detailed descriptions of image content, such as "two people shaking hands," or "people looking at each other," to each image's memorability map, in an effort to find out more about what makes the image memorable.
(摘自http://www.cnblogs.com/tabatabaye/articles/891241.html) 作图像处理方面的研究工作,最重要的两个问题:其一是要把握住国际上最前 沿的内容;其二是所作工作要具备很高的实用背景。解决第一个问题的办法就 是找出这个方向公认最牛的几个超级大拿(看看他们都在作什么)和最权威的 出版物(阅读上面最新的文献),解决第二个问题的办法是你最好能够找到一个 实际应用的项目,边做边写文章。 做好这几点的途径之一就是充分利用网络资源,特别是权威网站和大拿们的个人主页。下面是我收集的一些资源,希望对大家有用。(这里我要感谢SMTH AI版的alamarik和Graphics版的faintt) 导航栏: 研究群体 大拿主页 前沿期刊 GPL软件资源 搜索引擎 一、研究群体 http://www-2.cs.cmu.edu/~cil/vision.html 这是卡奈基梅隆大学的计算机视觉研究组的主页,上面提供很全的资料,从发表文章的下载到演示程序、测试图像、常用链接、相关软硬件,甚至还有一个搜索引擎。 http://www.cmis.csiro.au/IAP/zimage.htm 这是一个侧重图像分析的站点,一般。但是提供一个Image Analysis环境---ZIMAGE and SZIMAGE。 http://www.via.cornell.edu/ 康奈尔大学的计算机视觉和图像分析研究组,好像是电子和计算机工程系的。侧重医学方面的研究,但是在上面有相当不错资源,关键是它正在建设中,能够跟踪一些信息。 http://www2.parc.com/istl/groups/did/didoverview.shtml 有一个很有意思的项目:DID(文档图像解码)。 http://www-cs-students.stanford.edu/ 斯坦福大学计算机系主页,自己找吧:( http://www.fmrib.ox.ac.uk/analysis/ 主要研究:Brain Extraction Tool,Nonlinear noise reduction,Linear Image Registration, Automated Segmentation,Structural brain change analysis,motion correction,etc. http://www.cse.msu.edu/prip/ 这是密歇根州立大学计算机和电子工程系的模式识别--图像处理研究组,它的FTP上有许多的文章(NEW)。 http://pandora.inf.uni-jena.de/p/e/index.html 德国的一个数字图像处理研究小组,在其上面能找到一些不错的链接资源。 http://www-staff.it.uts.edu.au/~sean/CVCC.dir/home.html CVIP(used to be CVCC for Computer Vision and Cluster Computing) is a research group focusing on cluster-based computer vision within the Spiral Architecture. http://cfia.gmu.edu/ The mission of the Center for Image Analysis is to foster multi-disciplinary research in image, multimedia and related technologies by establishing links between academic institutes, industry and government agencies, and to transfer key technologies to help industry build next generation commercial and military imaging and multimedia systems. http://peipa.essex.ac.uk/info/groups.html 可以通过它来搜索全世界各地的知名的计算机视觉研究组(CV Groups),极力推荐。 二、图像处理GPL库 http://www.ph.tn.tudelft.nl/~klamer/cppima.html Cppima 是一个图像处理的C++函数库。这里有一个较全面介绍它的库函数的文档,当然你也可以下载压缩的GZIP包,里面包含TexInfo格式的文档。 http://iraf.noao.edu/ Welcome to the IRAF Homepage! IRAF is the Image Reduction and Analysis Facility, a general purpose software system for the reduction and analysis of astronomical data. http://entropy.brni-jhu.org/tnimage.html 一个非常不错的Unix系统的图像处理工具,看看它的截图。你可以在此基础上构建自己的专用图像处理工具包。 http://sourceforge.net/projects/ 这是GPL软件集散地,到这里找你想要得到的IP库吧。 三、搜索资源 当然这里基本的搜索引擎还是必须要依靠的,比如Google等,可以到我常用的链接看看。下面的链接可能会节省你一些时间: http://sal.kachinatech.com/ http://cheminfo.pku.edu.cn/mirrors/SAL/index.shtml 四、大拿网页 http://www.ai.mit.edu/people/wtf/ 这位可是MIT人工智能实验室的BILL FREEMAN。大名鼎鼎!专长是:理解--贝叶斯模型。 http://www.merl.com/people/brand/ MERL(Mitsubishi Electric Research Laboratory)中的擅长“Style Machine”高手。 http://research.microsoft.com/~ablake/ CV界极有声望的A.Blake 1977年毕业于剑桥大学三一学院并或数学与电子科学学士学位。之后在MIT,Edinburgh,Oxford先后组建过研究小组并成为Oxford的教授,直到1999年进入微软剑桥研究中心。主要工作领域是计算机视觉。 http://www-2.cs.cmu.edu/afs/cs.cmu.edu/user/har/Web/home.html 这位牛人好像正在学习汉语,并且搜集了诸如“两只老虎(Two Tigers)”的歌曲,嘿嘿:) 他的主页上面还有几个牛:Shumeet Baluja, Takeo Kanade。他们的Face Detection作的绝对是世界一流。他毕业于卡奈基梅隆大学的计算机科学系,兴趣是计算机视觉。 http://www.ifp.uiuc.edu/yrui_ifp_home/html/huang_frame.html 这位老牛在1963年就获得了MIT的博士学位!他领导的Image Lab比较出名的是指纹识别。 -------------------------------------------------------------------------------- 下面这些是我搜集的牛群(大部分是如日中天的Ph.D们),可以学习的是他们的Study Ways! Finn Lindgren(Sweden):Statistical image analysis http://www.maths.lth.se/matstat/staff/finn/ Pavel Paclik(Prague):statistical pattern recognition http://www.ph.tn.tudelft.nl/~pavel/ Dr. Mark Burge:machine learning and graph theory http://cs.armstrong.edu/burge/ yalin Wang:Document Image Analysis http://students.washington.edu/~ylwang/ Geir Storvik: Image analysis http://www.math.uio.no/~geirs/ Heidorn http://alexia.lis.uiuc.edu/~heidorn/ Joakim Lindblad:Digital Image Cytometry http://www.cb.uu.se/~joakim/index_eng.html S.Lavirotte: http://www-sop.inria.fr/cafe/Stephane.Lavirotte/ Sporring:scale-space techniques http://www.lab3d.odont.ku.dk/~sporring/ Mark Jenkinson:Reduction of MR Artefacts http://www.fmrib.ox.ac.uk/~mark/ Justin K. Romberg:digital signal processing http://www-dsp.rice.edu/~jrom/ Fauqueur:Image retrieval by regions of interest http://www-rocq.inria.fr/~fauqueur/ James J. Nolan:Computer Vision http://cs.gmu.edu/~jnolan/ Daniel X. Pape:Information http://www.bucho.org/~dpape/ Drew Pilant:remote sensing technology http://www.geo.mtu.edu/~anpilant/index.html 五、前沿期刊(TOP10) 这里的期刊大部分都可以通过上面的大拿们的主页间接找到,在这列出主要是为了节省直接想找期刊投稿的兄弟的时间:) IEEE Trans. On PAMI http://www.computer.org/tpami/index.htm IEEE Transactionson Image Processing http://www.ieee.org/organizations/pubs/transactions/tip.htm Pattern Recognition http://www.elsevier.com/locate/issn/00313203 Pattern Recognition Letters http://www.elsevier.com/locate/issn/01678655
The Integral Image or Summed Area Table , was first introduced to us in 1984, but wasn’t properly introduced to the world of Computer Vision till 2001 by Viola and Jones with the Viola-Jones Object Detection Framework . The Integral Image is used as a quick and effective way of calculating the sum of values (pixel values) in a given image – or a rectangular subset of a grid (the given image). It can also, or is mainly, used for calculating the average intensity within a given image. If one wants to use the Integral Image, it is normally a wise idea to make sure the image is in greyscale first. How does it work? So, how does it work? I hear you cry. Well, in all honesty it really isn’t that hard to get the grasp of! Sure some of the equations may look daunting, but they really aren’t that hard… So, let us start off with the basics. When creating an Integral Image, we need to create a Summed Area Table. In this table, if we go to any point (x,y) then at this table entry we will come across a value. This value itself is quite interesting, as it is the sum of all the pixel values above, to the left and of course including the original pixel value of (x,y) itself: What is really good about the Summed Area Table, is that we are actually able to construct it with only one pass over of the given image. There will be more on complexity later, but, in order for this fact to become true, all we have to do is accept that the value in the Summed Area Table at (x,y) is simply calculated by: That is, we get the original pixel value i(x,y) form the image, and then we add the values directly above this pixel, and directly left to this pixel from the Summed Area Table at s(x-1, y) and s(x, y-1). Finally, we subtract the value directly top-left of i(x,y) from the Summed Area Table – that is, s(x-1, y-1). Here is a better example, take the following image and its corresponding Summed Area Table: On the left we have the given image, with its corresponding pixel values. On the right we have the images corresponding Summed Area Table. AT this current moment in time, we only have one value filled in, that is to say, we have filled in s(x-1, y-1) = 5. Why is this? Well, taking the equation from above. We simply substitute in values: i(x,y) = 5 - this is the pixel value from the given image. The next values are from the Summed Area Table. s(x-1, y) = 0 - why? because x-1 is outside the image bounds, so it is automatically a value of 0. s(x, y-1) = 0 – can you see why this one is? That’s right, same as above, but this time because of y-1. s(x-1, y-1) = 0 - this one should be obvious by now. In the case of all s(x’, y’) above, they were all out-of-bounds inside the Summed Area Table. So are all defaulted to a value of 0. Therefore, 5+0+0-0 = 5. So s(x-1, y-1) gets a value of 5. Assuming here of course that s(x-1, y-1) is s(x,y) for the time being for the equation above. Now, I am just going to substitute in the values for s(x, y-1) and s(x-1, y). It is up to YOU to check them, and see if they’re are correct – as well as to see if you can use the equation correctly: If you have actually attempted to calculate the values, and ended up with the correct ones. Well done! You may give yourselves a pat on the back! This is what those values represent. taking the s(x-1, y) entry, we have a value of 8. This value represents the sum of the pixels to the left, above and including itself. In this case we have the pixel itself with value 3, and using the equation, we use the entry in the Summed Area Table directly above (only none out-of-bounds result), which has a value of 5. So 5+3=8, which IS the sum of the pixels left, above and including itself. But, now I will show you quickly the calculation for s(x,y) using 4 values. If you struggled slightly with trying to solve the values for s(x-1, y) etc. then this should help you a little bit; otherwise, feel free to drop off a comment with any questions. Here, is the Summed Area Table completed: Some of you may have already calculated it, but here’s what you do. First of all, we sub in the values from the above tables: i(x,y) = 6 - Remember, this value comes from the actual given image. Which is as marked, 6. s(x-1, y) = 8 - This time we need the values in the Summed Area Table. this value here is 8. s(x, y-1) = 7 - Same as above, so this time the value is 7. s(x-1, y-1) = 5 - This time we can actually use this value, which is a value of 5 This time all values can be used, as there are no out-of-bounds results. So now, sticking these values into the equation we get: s(x,y) = 6 + 8 + 7 – 5 = 16 . The question is, is this correct? Well yes, it is! Remembering that 16 is the sum of all pixels top, left and itself, we add up the 4 pixel values in the actual given image: 5 + 2 + 3 + 6 = 16 !! Amazing isn’t it! What next? Well, once you have used the equation to calculate and fill up your Summed Area Table, the the task of calculating the sum of pixels in some rectangle which is a subset of the original image can be done in constant time. Yes, that’s right, in O(1) complexity! In order to do this we only need to use 4 values from the Summed Area Table, that is, 4 array references into the Summed Area Table. With these 4 values, we then add or subtract them for the correct value of the sum of the pixels within that region. To do this, we use this equation: This is fairly similar to the equation further above. Let us now have an example. Let us say that we wish to calculate the area contained by the green square: Remember, the value 16 is the total sum of all the squares. But we want just that green square. As you can already see I have labelled on the A, B, C and D. This is what each of them are: Firstly, we have s(A), which includes these squares: So s(A) is just the green square, which has value 5. Next we have s(B): Now, s(B) is the value 7, because this is the value of the sum of the values up to that point. s(C) looks fairly similar: As for s(D), this is the sum of all the values up to that point: So, with all of this, we have the values: A = 5 B = 7 C = 8 D = 16 With this, we can substitute them all into the equation for, i(x’, y’) = s(A) + s(D) – s(B) – s(C) = 5 + 16 – 7 – 8 = 6 Therefore, we are left with the value 6. Think it’s wrong? Think again! Go back to the original image pixel values. Now look at the bottom-right pixel value. What’s that, a 6! See told you it worked! Bigger Example I am not going to explain the whole process for this, it’s up to you to work it out. But here is an example of a bigger original image (4×4), with its corresponding Summed Area Table. The final 5 images are for calculating that area enclosed in A, B,C and D labels. Original Image Summed Area Table Calculating an Area This is the area we want. This is what A, B, C and D correspond to. Remember to use A+D-B-C . If you do everything correctly you should get the value 16 . Complexity We have already touched a little bit on this. But as mentioned previously, the complexity for evaluating the Summed Area Table can be done in O(1) rather than in O(n^2) . Notice : as time goes on this post will probably get further improved. 转载自: http://computersciencesource.wordpress.com/2010/09/03/computer-vision-the-integral-image/
The benefits of multisensor fusion have motivated research in this area in recent years. Redundant fusion methods are used to enhance fusion system capability and reliability. The benefits of beyond wavelets have also prompted scholars to conduct research in this field. In this paper, we propose the maximum local energy method to calculate the low-frequency coefficients of images and compare the results with those of different beyond wavelets. An image fusion step was performed as follows: first, we obtained the coefficients of two different types of images through beyond wavelet transform. Second, we selected the low-frequency coefficients by maximum local energy and obtaining the high-frequency coefficients using the sum modified Laplacian method. Finally, the fused image was obtained by performing an inverse beyond wavelet transform. In addition to human vision analysis, the images were also compared through quantitative analysis. Three types of images (multifocus, multimodal medical, and remote sensing images) were used in the experiments to compare the results among the beyond wavelets. The numerical experiments reveal that maximum local energy is a new strategy for attaining image fusion with satisfactory performance.
Name SCI/SCIE ImpactFactor(2010) WebSite TPAMI SCISCIE 4.378 http://science.thomsonreuters.com/cgi-bin/jrnlst/jlcovchanges.cgi?PC=D IJCV SCISCIE 3.508 http://www.springer.com/computer/image+processing/journal/11263 SMC-B SCISCIE 3.007 http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=3477 IEEE T ImageProcessing SCISCIE 2.848 http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=83 PR SCISCIE 2.554 http://www.elsevier.com/locate/pr IEEE T Signal Processing SCISCIE 2.212 http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=78 ComputerVisionImageUnderstanding SCISCIE 1.676 http://www.elsevier.com/locate/cviu ImageVisionComputing SCISCIE 1.474 http://www.elsevier.com/locate/imavis PatternRecognitionLetters SCIE 1.303 http://www.elsevier.com/locate/patrec IEEE SignalProcessingLetters SCIE 1.173 http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=97 Signal Processing SCIE 1.135 http://www.elsevier.com/locate/sigpro MachineVisionApplications SCIE 0.952 http://www.scimagojr.com/journalsearch.php?q=12984tip=sidclean=0 IET Image Processing SCIE 0.758 http://scitation.aip.org/IET-IPR IET Signal Processing SCIE 0.794 http://scitation.aip.org/IET-SPR Journal of Electronic Imaging SCIE 0.444 http://spie.org/x868.xml IEEJ Transactions on Electrical and Electronic Engineering SCIE 0.361 http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1931-4981 JournalComputer Scienceechnology SCIE 0.632 http://jcst.ict.ac.cn:8080/jcst/EN/volumn/home.shtml Chinese Journal of Electronics SCIE 0.156 http://www.ejournal.org.cn/english/en/qkjs.asp Journal of Infrared Millimeter Waves SCISCIE Journal of Infrared , Millimeter , and Terahertz Waves SCISCIE 1 IEEE Transactions on Image Processing 2 IEEE Transactions on Multimedia 3 IEEE Transactions on Consumer Electronics 4 IEEE Signal Processing Magazine 5 IEEE Signal Processing Letters 6 IEEE Transactions on Signal Processing 7 IEEE Transactions on Instrumentation and Measurement 8 IEEE Journal of Selected Topics in Signal Processing 9 Electronics Letters 10 Journal of Imaging Science and Technology 11 Journal of Electronic Imaging 12 Image and Vision Computing 13 Signal Processing-Image Communication 14 Journal of Visual Communication and Image Representation 15 Computer Vision and Image Understanding 16 Signal Processing 17 EURASIP Journal on Image and Video Processing 18 SMPTE Motion Imaging Journal 19 Multimedia Tools and Applications 20 Journal of Signal Processing Systems 21 Acm transactions on graphics 22 ACM Transactions on Multimedia Computing Communications and Applications 23 Chinese Journal of Electronics
科院的博客可以上传pdf文件,挺好。以后把文章都上传上来 硕士期间(2004-0007),主要是从事激光扫描仪数据处理的工作,发表的主要文章如下: 1.基于激光扫描仪数据的建筑物立面特征信息提取 基于激光扫描仪数据的建筑物立面特征信息提取 2.基于激光扫描数据的建筑物信息格网化提取方法 基于激光扫描数据的建筑物信息格网化提取方法 3. 基于激光扫描回光强度的建筑物立面信息提取与分类 基于激光扫描回光强度的建筑物立面信息提取与分类 博士期间(2007-2010)论文,主要从事SAR图像处理、冰川运动速度估算方面的工作。 1. International Journal of Remote Sensing Feature-based image registration using the shape c ontext 2. IGARSS2009会议 Derivation of glacier velocity from SAR and optica l images with feature tracking 科院工作期间论文(2010-) 1.Remote Sensing of Environment Classification and snow line detection for glacial areas using the polarimetric SAR image.pdf 2.International Journal of Remote Sensing Comparison of SAR and optical data in deriving glacier velocity with feature tracking.pdf
怎样研究固体物理? 从水母bbs上看到了,在Science版精华区的一个八卦,作者kacmoody: 我想起了一段话。苏肇冰院士有次报告的时候说,他以前到科大向吴杭生请教t-J模型的推导,吴杭生把他骂了一顿,说他没读懂固体物理,固体物理学就是写下一个哈密顿量,然后做计算,然后把计算结果跟实验对比。 Zee曾经说起,有次Anderson去Santa Barara访问,对研究生们说,You do not write a Hamiltonian and do calculations, you just imagine what the electrons like to do. 呵呵,为什么Anderson拿了noble prize,吴杭生毕生只做了一个求超导转变温度的级数展开方法,真的不是偶然的,赫赫 最近我在学凝聚态物理,看到这样的八卦还是很有启发的。固体物理里面的哈密顿确实很麻烦,很难求解。我一点基础都没有。Anderson的这番话给了我一点信心。我现在需要做的就是闭上眼,想象光子晶体中的光子是怎么运动的,呵呵! 至于后面的评论讨论,挺专业的,所以不知所云。但是有两段可能对于如何image有所帮助: 安德逊境界高,地球人都知道,不过对后辈来说,闭上眼睛想也是要有基础的,比如不知道Maxwell Eqs,WF,光是想波或是想粒子其实是无用的。所以尽量不用数学,和尽量用数学来思考物理问题都是很好的锻炼。 想象是最有趣的事情。。对我这样的fresh guy是这么觉得 不过老师也说。。想是没错的 但是要算! 哎 我得到的教训是根据不足 还有基础不扎实 想的结果是越想越怀疑。。。很糟糕 一边算一边想了。。。目的是能够想清楚~ =======转载自 在格致上看到的,原作者zqyin http://gezhi.org/node/357 ============
1. Image processing with partial dierential equations 2. Image restoration via PDE's 3. Recent UCLA Computational and Applied Mathematics Reports . 4. IMAGERS UCLA Image Processing Group 5. Analysis PDE conferences 6. IEEE Transactions on Image Processing 7. Computer Science Journals 8. Reading group on medical image processing 9. Minerva Research Group 10. Penn Image Computing and Science Laboratory (PICSL) 11. Scientific Computing and Imaging (SCI) Institute and the School of Computing at the University of Utah 12. VISTA - Spatio-Temporal Vision