#
to create a panorama image.
input:
output:
前言：最近由于工作的關(guān)系，接觸到了很多篇以前都沒有聽說過的經(jīng)典文章，在感嘆這些文章偉大的同時(shí)，也頓感自己視野的狹小。 想在網(wǎng)上找找計(jì)算機(jī)視覺界的經(jīng)典文章匯總，一直沒有找到。失望之余，我決定自己總結(jié)一篇，希望對(duì) CV 領(lǐng)域的童鞋們有所幫助。由于自己的視野比較狹窄，肯定也有很多疏漏，權(quán)當(dāng)拋磚引玉了，如果你覺得哪篇文章是非常經(jīng)典的，也可以把相關(guān)信息連帶你的昵稱發(fā)給我，我好補(bǔ)上。我的信箱 xdyang.ustc@gmail.com
文章主要來源：PAMI, IJCV, TIP, CVIU, PR, IVC, CVGIU, CVPR, ICCV, ECCV, NIPS, SIGGRAPH, BMVC等
主要參考網(wǎng)站: Google scholar, citeseer, cvpapers, opencv 中英文官方網(wǎng)站
主要參考書籍：
數(shù)字圖像處理 第三版 岡薩雷斯等
圖像處理，分析和機(jī)器視覺 第三版 Sonka等（非常非常好的一本書）
學(xué)習(xí)OpenCV
計(jì)算機(jī)視覺：算法與應(yīng)用
文章按時(shí)間排序，排名不分先后，^_^。每一行最后一欄是我自己加的注釋，如果不喜歡可以無視之，如果有不對(duì)的地方還請(qǐng)告訴我，免得繼續(xù)出丑。 給出的文章有些是從google scholar或者citeseer上拷貝下來的，所以有鏈接。所有的文章在網(wǎng)上都很容易找到。有空的時(shí)候我會(huì)把它們?nèi)空沓鰜恚鸩缴蟼鞯絠share.iask.sina.com
由于整理的很倉促，時(shí)間也很短，還有很多不完善的地方。我會(huì)不斷改進(jìn)，并不時(shí)上傳新版本。
上傳地址為http://iask.sina.com.cn/u/2252291285/ish?folderid=775855
最后更新：2012/3/14
1990 年之前
| Peter Burt, Edward Adelson |
The Laplacian Pyramid as A Compact Image Code |
雖說這個(gè)Laplacian Pyramid是有冗余的，但使用起來非常簡(jiǎn)單方便，對(duì)理解小波變換也非常有幫助。這位Adelson是W.T.Freeman的老板，都是大牛. |
| J Canny |
A Computational Approach to Edge Detection |
經(jīng)典不需要解釋。在 Sonka的書里面對(duì)這個(gè)算法也有比較詳細(xì)的描述。 |
| S Mallat. |
A theory for multiresolution signal decomposition: The wavelet representation |
Mallat的代表作 |
| M Kass, A Witkin, D Terzopoulos. |
Snakes: active contour models |
Deformable model的開山鼻祖。 |
| RM HARALICK |
Textural Features for Image Classi?cation |
這三篇都是關(guān)于紋理特征的，雖然過去這么多年了，現(xiàn)在在檢索和識(shí)別中依然很有用。 |
| RM HARALICK |
Statistical and structural approaches |
|
| Tamura等 |
Texture features corresponding to visual perception |
|
| A P Dempster, N M Laird, D B Rubin. 1977 |
Maximum likelihood from incomplete data via the EM algorithm |
EM 算法在計(jì)算機(jī)視覺中有著非常重要的作用 |
| L Rabiner. 1989 |
A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition |
HMM 同樣是計(jì)算機(jī)視覺必須掌握的一項(xiàng)工具 |
| B D Lucas, T Kanade |
An iterative image registration technique with an application to stereo- vision |
Lucas 光流法 |
| J R Quinlan |
Induction of decision trees |
偏模式識(shí)別和機(jī)器學(xué)習(xí)一點(diǎn) |
1990 年
| P Perona, J Malik. PAMI |
Scale-space and edge detection using anisotropic diffusion |
關(guān)于 scale space 最早的一篇論文之一，引用率很高 |
| T Lindeberg |
Scale-space for discrete signals. |
Lindeberg 關(guān)于 scale space 比較早的一篇，后續(xù)還有好幾篇 |
| anzad, A.; Hong, Y.H. |
Invariant image recognition by Zernike moments |
Zernike moment,做過模式識(shí)別或者檢索的應(yīng)該都知道這個(gè)東東 |
1991 年
| W Freeman, E Adelson. |
The design and use of steerable filters |
Freeman最早的一篇力作，也是我讀的第一篇學(xué)術(shù)論文。現(xiàn)在Freeman在 MIT 風(fēng)生水起，早已是IEEE Fellow了 |
| Michael J. Swain , Dana H. Ballard |
Color Indexing. |
google scholar 上引用將近五千次 |
| MA Turk CVPR |
Face recognition using eigenfaces |
|
1992 年
| L G Brown. |
A survey of image registration techniques. |
比較早的一篇關(guān)于配準(zhǔn)的綜述了 |
1993 年
| S G Mallat, Z Zhang. |
Matching pursuits with time-frequency dictionaries |
Mallat另一篇關(guān)于小波的文章，不研究小波的可以無視之 |
| L Vincent. |
Morphological grayscale reconstruction in image analysis: Applications and efficient algorithms |
|
| DP Huttenlocher |
Comparing images using the Hausdorff distance |
Google scolar 上引用2200多次 |
1994 年
| J Shi, C Tomasi. |
Good feature to track. |
Tomasi這個(gè)名字還會(huì)出現(xiàn)好幾次，真的很牛 |
| Linderberg |
Scale-space theory in computer vision |
|
| J L Barron, D J Fleet, S S Beauchemin. |
Performance of optical flow techniques. |
|
1995 年
| R Malladi, J Sethian, B Vemuri. |
Shape Modeling with Front Propagation: A Level Set Approach |
Level set的經(jīng)典文章 |
| TF COOTES |
Active Shape Models-Their Training and Application |
ASM |
| MA Stricker |
Similarity of color images |
顏色檢索相關(guān) |
| C Cortes, V Vapnik. |
Support-vector networks. |
SVM 在計(jì)算機(jī)視覺中也有著非常重要的地位 |
1996 年
| T MCINERNEY. |
Deformable models in medical image analysis: A survey |
活動(dòng)模型的一篇較早的綜述 |
| Tai Sing Lee |
Image Representation Using 2D Gabor Wavelets |
Google引用也有近千次 |
| Amir Said, A. Pearlman |
A New, Fast, and Ef?cient Image Codec Based on Set Partitioning in Hierarchical Tree |
SPIHT。圖像壓縮領(lǐng)域與 EBCOT齊名的經(jīng)典算法。 |
| L P Kaelbling, M L Littman, A W Moore. |
Reinforcement learning: A survey |
機(jī)器學(xué)習(xí)里面的一篇綜述，引用率比較高，就列在這了。 |
| B. S. Manjunath and W. Y. Ma |
Texture features for browsing and retrieval of image data |
檢索的文章比較多，其實(shí)它們的應(yīng)用不僅僅是檢索。只要是需要提取特征的地方，檢索里面的方法都可以用到 |
| |
comparing images using color coherence vectors |
檢索中的CCV方法 |
| |
Image retrieval using color and shape |
關(guān)于形狀特征后面有一篇綜述 |
1997 年
| V Caselles, R Kimmel, G Sapiro. |
Geodesic active contours |
活動(dòng)輪廓模型的一個(gè)小分支 |
| R E Schapire, Y Freund, P Bartlett, W S Lee. |
Boosting the Margin: A New Explanation for the Effectiveness of Voting Methods. |
Schapire 和 Freund 發(fā) 明 了Adaboost，給計(jì)算機(jī)視覺帶來了不少經(jīng)典算法 |
| F Maes, D Vandermeulen, G Marchal, P Suetens. |
Multimodality image registration by maximization of mutual information |
互信息量配準(zhǔn) |
| E Osuna, R Freund, F Girosi. |
Training support vector machines: An application to face detection. |
SVM在人臉檢測(cè)中的應(yīng)用。不過人臉檢測(cè)最經(jīng)典的方法應(yīng) 該是Viola-Jones |
| J Huang, S Kumar, M Mitra, W-J Zhu, R Zabih. |
Image indexing using color correlogram |
Color correlogram，檢索中的又一個(gè)顏色特征。和前面的 CCV 以及顏色矩特征基本上覆蓋了所有的顏色特征。 |
| Y Freund, R Schapire. |
A decisiontheoretic generalization of on-line learning and an application to boosting. |
Adaboost的經(jīng)典文章 |
1998 年
1998 年是圖像處理和計(jì)算機(jī)視覺經(jīng)典文章井噴的一年。大概從這一年開始，開始有了新的趨勢(shì)。由于競(jìng)爭(zhēng)的加劇，一些好的算法都先發(fā)在會(huì)議上了，先占個(gè)坑，等過一兩年之后再擴(kuò)展到會(huì)議上。
| T Lindeberg |
Feature detection with automatic scale selection |
Linderberg的 scale space到此為止基本結(jié)束了。在一些邊緣提取，道路或者血管檢測(cè)中，scale space 確實(shí)是一種很不錯(cuò)的工具 |
| C J C Burges. |
A tutorial on support vector machines for pattern recognition. |
使用 svm的話，這篇文章應(yīng)該是必讀的了。比 95 年那篇原始文章引用率還高 |
| M Isard, A Blake. |
CONDENSATION – Conditional TrackingDensity Propagation for Visual |
Tracking中的經(jīng)典文章了 |
| L Page, S Brin, R Motwani, T Winograd |
The PageRank citation ranking: bringing order to the web |
這篇文章應(yīng)該不屬于 CV 的范疇，鑒于作者的大名鼎鼎，暫且列在這 |
| C Tomasi, R Manduchi. |
Bilateral filtering for gray and color images. |
做過圖像濾波平滑去噪或者 HDR的應(yīng)該都知道Bilateral filter。原理非常非常簡(jiǎn)單，簡(jiǎn)單到一個(gè)公式就可以概括這篇文章，簡(jiǎn)單到實(shí)在無法擴(kuò)充到期刊。這也是 Tomasi 第二次出現(xiàn)了。一直很納悶，這個(gè)很直觀的思想在這之前怎么就從來沒人提呢。 |
| C Xu, J L Prince. |
Snakes, shapes and gradient vector flow. |
終于碰到中國人寫的文章了，很榮幸還是校友。GVF是 snake和levelset領(lǐng)域的重要分支和方法 |
| Wim Sweldens. |
The lifting scheme: A construction of second generation wavelets. |
第二代小波。真正讓小波有了實(shí)用價(jià)值，在 JPEG2000 中就采用的提升小波。個(gè)人更喜歡的是下一篇，簡(jiǎn)單易懂，字體也大 |
| Daubechies Wim Sweldens |
Factoring wavelet transforms into lifting steps |
另一位作者也很牛，小波十講的作者 |
| H A Rowley, S Baluja, T Kanade. |
Neural Network-based Face Detection. |
做人臉的應(yīng)該是必看的了。不做人臉的話應(yīng)該可以不用看吧 |
| J B A Maintz, M A Viergever. |
A survey of medical image registration. |
關(guān)于圖像配準(zhǔn)的另一篇綜述 |
| T F Cootes, G J Edwards, C J Taylor. |
Active Appearance Models |
AAM |
1999 年
| D Lowe. |
Object Recognition from Local Scale-invariant Features |
大名鼎鼎的SIFT，后面有一篇IJCV上的 Journal版本，更全面一點(diǎn)。 |
| R E Schapire. |
A brief Introduction to Boosting |
還是 boosting |
| D M Gavrila. |
The visual analysis of human movements: a survey |
綜述文章的引用一般都比較高 |
| Y Rui, T S Huang, S F Change. |
Image retrieval: current techniques, promising directions, and open issues |
TSHuang小組對(duì)檢索的一個(gè)總結(jié) |
| J K Aggarwal, Q Cai. |
Human motion analysis: a review |
人體運(yùn)動(dòng)分析的一個(gè)綜述 |
2000 年
世紀(jì)之交，各種綜述都出來了
| J Shi, J Malik. |
Normalized Cuts and Image Segmentation |
NCuts的引用率相當(dāng)高，Jianbo Shi也因?yàn)檫@篇文章成為計(jì)算機(jī)視覺界引用率最高的作者之一 |
| Z Zhang. |
A Flexible New Technique for Camera Calibration |
張正友的關(guān)于攝像機(jī)標(biāo)定的經(jīng)典短文 |
| A K Jain, R P W Duin, J C Mao. |
Statistical pattern recognition: a review. |
統(tǒng)計(jì)模式識(shí)別綜述，這一年 pami上兩篇很有名的綜述之一。 在這里推薦 Web 寫的 Statistical Pattern Recognition第三版，相當(dāng)不錯(cuò)，網(wǎng)上有電子版。 |
| C Stauffe |
Learning Patterns of Activity Using Real-Time Tracking |
搜 TLD 的時(shí)候發(fā)現(xiàn)這篇文章引用率也很高，兩千多次。還沒來得及讀。 |
| D Taubman. |
High performance Scalable Image Compression With EBCOT |
EBCOT，JPEG2000 中的算法 |
| A W M Smeulders, M Worring, S Santini, A Gupta, R Jain. |
Content-based image retrieval at the end of the early years |
在世紀(jì)之交對(duì)圖像檢索的一篇很權(quán)威的綜述。感覺在這之后檢索的研究也沒那么熱了。不過在工業(yè)界熱度依舊，各大網(wǎng)上購物平臺(tái)，比如淘寶， 亞馬遜，京東等都在做這方面的研發(fā)，衣服檢索是一個(gè)很不錯(cuò)的應(yīng)用點(diǎn)。 |
| M Pantic, L J M Rothkrantz. |
Automatic analysis of facial expressions: the state of the art. |
|
| N Paragios, R Deriche. |
Geodesic active contours and level sets for the detection and tracking of moving objects |
使用 level set做跟蹤 |
| Y Rubner, C Tomasi, L Guibas. |
TThe earth mover’s distance as a metric for image retrieval. |
EMD算法。Tomasi再次出現(xiàn) |
| |
PicToSeek Combining Color and Shape Invariant Features for Image Retrieval |
依然是檢索特征 |
2001 年
| Paul Viola, Michael J Jones. |
Robust real-time object detection |
這是一篇很牛的文章，在人臉檢測(cè)上幾乎成了標(biāo)準(zhǔn)。比較坑爹的是，號(hào)稱發(fā)在IJCV2001 上，但怎么找也找不到。應(yīng)該是 IJCV2004年的那篇“Robust real-time face detection”吧。 他們?cè)谶@一年另一篇比較出名的文章是在CVPR上的“Rapid ObjectDetection using a Boosted Cascade of Simple Features”這篇才是04年那篇著名文章的會(huì)議版。 |
| Y Boykov, Kolmogorov. |
An experimental comparison of min-cut/max-flow algorithms for energy minimization in vision. |
俄羅斯人在 graph cut 領(lǐng)域開始發(fā)力了 |
| T Moeslund, E Granum. |
A Survey of Computer Vision Based Human Motion Capture |
人體運(yùn)動(dòng)綜述 |
| T F Chan, L Vese. |
Active contours without edges. |
Snake 和 level set領(lǐng)域的經(jīng)典文章 |
| A M Martinez, A C Kak. |
PCA versus LDA |
PCA 也是計(jì)算機(jī)視覺中非掌握不可的工具。LDA在模式識(shí)別中有很重要的地位 |
| BS Manjunath |
Color and texture descriptors |
顏色和紋理的描述子，在識(shí)別中很有用 |
2002 年
| D Comaniciu, P Meer. |
Mean shift: A robust approach toward feature space analysis. |
Mean shift的經(jīng)典文章。前兩天發(fā)現(xiàn) Comaniciu 已經(jīng)是 IEEE Fellow了 |
| Ming-Husan Yang, David J Kriegman, Narendra Ahuja. |
Detecting Faces in Images: A Survey. |
人臉檢測(cè)綜述，引用率想不高都難 |
| R Hsu, M Abdel-Mottaleb. |
Face Detection in Color Images. |
依然是人臉檢測(cè)，名字都起得這么霸氣 |
| J-L Starck, E J Candès, D L Donoho. |
The curvelet transform for image denoising. |
Geometrical wavelet 中的一篇代表 作 。 其 他 的 如 ridgelet, contourlet, bandelet 等在這里就不贅述了。研究這方面的很容易找到這方面的經(jīng)典文獻(xiàn)。個(gè)人以為不研究這方面的看了后對(duì)自己的研究也不會(huì)有多大啟發(fā)。曾經(jīng)以為這個(gè)方向會(huì)很火，到最后還是沒火起來。 我覺得原因可能是現(xiàn)在存儲(chǔ)和傳輸能力的大大提高，使得對(duì)壓縮的需求沒有那么大了，這方面的研究自然就停滯了，就如同JPEG2000沒有成氣候 |
| |
Shape matching and object recognition using shape contexts |
Shape context。用形狀匹配達(dá)到目標(biāo)識(shí)別目的。這方面最經(jīng)典的文章了。隨后后續(xù)也有一些這方面的文章，但基本都是很小的改進(jìn)或者應(yīng)用。作者提供了原碼，可以在 matlab上運(yùn)行看看效果。 |
| N Paragios, R Deriche. |
Geodesic active regions and level set methods for supervised texture segmentation |
|
| |
Statistical Color Models with Application to Skin Detection |
|
| |
A tutorial on particle filters for online nonlinear non-Gaussian Bayesian tracking |
particle filter 的一個(gè)綜述 |
2003 年
| W Zhao, R Chellappa, P J Phillips, A Rosenfeld. |
Face recognition: A literature survey. |
人臉檢測(cè)的綜述 |
| J Sivic, A Zisserman. |
Video Google: A text retrieval approach to object matching in videos. |
好像是Visual words的起源文章。引用率很高，先列出來再看。 |
| D Comaniciu, V Ramesch, P Meer. |
Kernel-Based Object Tracking. |
基于核的跟蹤。 |
| B Zitová, J Flusser. |
Image registration methods: A survey. |
又一篇圖像配準(zhǔn)的綜述。做圖像配準(zhǔn)的比較有福氣，綜述很多 |
| K Mikolajczyk, C Schmid. |
A performance evaluation of local descriptors. |
比較各種描述子的，包括SIFT |
| M J Wainwright, M I Jordan. |
Graphical models, exponential families, and variational inference. |
喬丹的名氣太大，不露露臉說不過去 |
| J Portilla, V Strela, M Wainwright, E Simoncelli. |
Image denoising using scale mixtures of gaussians in the wavelet domain. |
圖像去噪，小波變換，混合高斯 |
| Robert E. Schapire |
The Boosting Approach to Machine Learning An Overview |
boosting作者自己寫的綜述，自然值得一看。 |
2004 年
| |
Lucas-Kanade 20 Years On A Unifying Framework |
引用文章摘要的第一句話Since the Lucas-Kanade algorithm was proposed in 1981 image alignment has become one of the most widely used techniques in computer vision. Applications range from optical flow and tracking to layered motion, mosaic construction, and face coding. |
| D G Lowe. |
Distinctive image features from scale-invariant keypoints. |
SIFT，不解釋 |
| Chih-ChungChang,Chih-Jen Lin. |
LIBSVM: A library for support vectormachines |
我實(shí)在懷疑引用這篇文章的人是否都看過這篇文章。貌似不看這篇文章也可以使用 LIBSVM |
| Z Wang, A C Bovik, H R Sheikh, E P Simoncelli. |
Image quality assessment: From error visibility to structural similarity |
圖像質(zhì)量評(píng)價(jià)，最近 Bovik 還有一篇類似的文章也刊登在 TIP上 |
| Y Ke, R Sukthankar. |
Pca-sift: a more distinctive representation for local image descriptors |
SIFT 的變形 |
| |
Review of shape representation and description techniques |
|
| |
Efficient Graph-Based Image Segmentation |
|
2005 年
| N Dalal, B Triggs. |
Histograms of oriented gradients for human detection. |
HOG 雖然很新，但很經(jīng)典 |
| A C Berg, T L Berg, J Malik. |
Shape matching and object recognition using low distortion correspondences. |
還是 shape matching |
| S Roth, M Black. |
Fields of experts: A framework for learning image priors. |
這篇應(yīng)該要?dú)w結(jié)到圖像統(tǒng)計(jì)特性的范疇吧 |
| Z Tu, X Chen,A L Yuille, S C Zhu. |
Image parsing: Unifying segmentation, detection, and recognition. |
|
| |
Geodesic active regions and level set methods for motion estimation and tracking |
|
| Chunming Li, Chenyang Xu, Changfeng Gui, and Martin D. Fox |
Level Set Evolution Without Re-initialization: A New Variational Formulation |
這篇文章解決了level set中需要不停的重初始化的問題。在 2010 年的 TIP上有一篇 Journal版本 Distance Regularized Level Set Evolution and its Application to Image Segmentation |
| |
A Performance Evaluation of Local Descriptors |
前面那篇是會(huì)議的，這篇是 PAMI上的。比較各種描述子的，包括SIFT |
2006 年
| D Donoho. |
Compressed sensing. |
CS 壓縮感知 最近很火的一個(gè)名詞 |
| Greg Welch, Gary Bishop. |
An introduction to the Kalman Filter. |
kalman濾波 |
| S Lazebnik, C Schmid, J Ponce. |
Beyond bags of features: spatial pyramid matching for recognizing natural scene categories. |
Visual words |
| Xiaojin Zhu. |
Semi-supervised learning literature survey. |
|
| A Yilmaz, O Javed, M Shah. |
Object Tracking: A survey. |
tracking的一篇綜述 |
| |
Image Alignment and Stitching: A Tutorial |
|
2007 年
| |
A Review of Statistical Approaches to Level Set Segmentation: Integrating Color, Texture, Motion and Shape |
|
| |
The Appearance of Human Skin: A Survey |
|
| |
Local Invariant Feature Detectors: A Survey |
|
2008 年
| H Bay, A Ess, T Tuytelaars, L V Gool. |
SURF: Speeded Up Robust Features. |
|
| K E A van de Sande, T Gevers, C G M Snoek. |
Evaluation of Color Descriptors for Object and Scene Recognition |
|
| M Yang |
A Survey of Shape Feature Extraction Techniques |
雖然這篇文章的引用率目前來看并不高,但個(gè)人認(rèn)為這是一篇在shape feature方面很不錯(cuò)的文章 |
| P.Felzenszwalb, D. McAllester, D. Ramanan |
A Discriminatively Trained, Multiscale, Deformable Part Model |
2008 年的 CVPR，到現(xiàn)在引用已有四百多次，潛力巨大。rosepink提供 |
2009 年
| J Wright, A Y Yang, A Ganesh, S S Sastry, Ma. |
Robust Face Recognition via Sparse Representation. |
|
| B Settles. |
Active learning literature survey |
|
2010 年
2011 年
| |
Hough Forests for Object Detection, Tracking, and Action Recognition |
|
| |
Robust Principal Component Analysis? |
Candes 和 UIUC 的Ma Yi等人 |
2012 年
| Zdenek Kalal, Krystian Mikolajczyk,and Jiri Matas, |
Tracking-Learning-Detection |
PAMI上的，雖然還沒有正式發(fā)表，但肯定會(huì)火。在作者的主頁上有幾篇相關(guān)的會(huì)議文章， demo和code。用到了 Lucas-Kanade方法 |
| |
|
|
這些你都知道嗎？以后給老外作報(bào)告的時(shí)候可千萬別再“The equation is like this…”, “The value is like this…”
有些公式不好寫，能看出來什么意思就行了。
1.Logic
∃there exist
∀for all
p⇒q p implies q / if p, then q
p⇔q p if and only if q /p is equivalent to q / p and q are equivalent
2.Sets
x∈A x belongs to A / x is an element (or a member) of A
x∉A x does not belong to A / x is not an element (or a member) of A
A⊂B A is contained in B / A is a subset of B
A⊃B A contains B / B is a subset of A
A∩B A cap B / A meet B / A intersection B
A∪B A cup B / A join B / A union B
A\B A minus B / the diference between A and B
A×B A cross B / the cartesian product of A and B
3. Real numbers
x+1 x plus one
x-1 x minus one
x±1 x plus or minus one
xy xy / x multiplied by y
(x-y)(x+y) x minus y, x plus y
= the equals sign
x=5 x equals 5 / x is equal to 5
x≠5 x (is) not equal to 5
x≡y x is equivalent to (or identical with) y
x>y x is greater than y
x≥y x is greater than or equal to y
x<y x is less than y
x≤y x is less than or equal to y
0<x<1 zero is less than x is less than 1
0≤x≤1 zero is less than or equal to x is less than or equal to 1
|x| mod x / modulus x
x2 x squared / x (raised) to the power 2
x3 x cubed
x4 x to the fourth / x to the power 4
xn x to the nth / x to the power n
x (−n) x to the (power) minus n
x的平方根(square) root x / the square root of x
x的三次根cube root (of) x
x的四次根fourth root (of) x
x的n次根nth root (of) x
(x+y)2 x plus y all squared
n! n factorial
x^x hat
x¯ x bar
x˜ x tilde
xi xi / x subscript i / x suffix i / x sub i
∑(i=1~n) ai the sum from i equals one to n ai / the sum as i runs from 1 to n of the ai
4. Linear algebra
‖x‖the norm (or modulus) of x
OA→OA / vector OA
OA¯ OA / the length of the segment OA
AT A transpose / the transpose of A
A−1 A inverse / the inverse of A
5. Functions
f(x) fx / f of x / the function f of x
f:S→T a function f from S to T
x→y x maps to y / x is sent (or mapped) to y
f’(x) f prime x / f dash x / the (first) derivative of f with respect to x
f”(x) f double-prime x / f double-dash x / the second derivative of f with respect to x
f”’(x) triple-prime x / f triple-dash x / the third derivative of f with respect to x
f (4) (x) f four x / the fourth derivative of f with respect to x
∂f/∂x1 the partial (derivative) of f with respect to x1
∂2f/∂x12 the second partial (derivative) of f with respect to x1
∫0∞ the integral from zero to infinity
limx→0the limit as x approaches zero
limx→0+the limit as x approaches zero from above
limx→0−the limit as x approaches zero from below
logey log y to the base e / log to the base e of y / natural log (of) y
lny log y to the base e / log to the base e of y / natural log (of) y
摘要: computer vision(cv) 存在ICCV/CVPR/ECCV三個(gè)頂級(jí)會(huì)議, 它們檔次差不多,都應(yīng)該在一流會(huì)議行列, 沒有必要給個(gè)高下. 有些us的人認(rèn)為ICCV/CVPR略好于ECCV,而歐洲人大都認(rèn)為ICCV/ECCV略好于CVPR, 某些英國的人甚至認(rèn)為BMVC好于CVPR.簡(jiǎn)言之, 三個(gè)會(huì)議差不多, 各有側(cè)重和偏好.三者乃cv領(lǐng)域的旗艦和風(fēng)向標(biāo),其oral paper (包括be...
閱讀全文
1.打開音頻，從頭開始，向后聽。
2.一次聽一小段，建議聽5秒左右的長度。如果聽出來了，換句話說，也就是知道是哪些單詞，就寫下來。
3.如果不知道是哪些單詞，就再聽最多5遍，如果能聽出來，就寫下來，如果聽不出來，就查一下金山詞霸，看看模糊查詢里有沒有正確答案。如果還沒有，那么就先聽后面的內(nèi)容，等到整套聽寫都結(jié)束了之后，再看原文。如此循環(huán)往復(fù)，聽完原文。
4.聽寫結(jié)束之后，用軟件將自己聽寫的文本與原文進(jìn)行對(duì)比，找出不同點(diǎn)。
5.將那些自己與原文不一致的地方，全都重新聽5遍，保證這個(gè)單詞的發(fā)音記在自己的腦子里。
TPO24 1
in my opinion, the popular palce is our national park.
for starters, there is a large square in the park so it is convenient for people to do outdoor activities.
For example, I always see people exercising themselves there.
The icing on the cake is there is also a lake in the park and the lake is really beautiful. It is really a good place to relax your self when you get tired in daily life.
So this is the place people like to go and the reason why people enjoy going there.
propose
proposal
TPO24 4
In the reading material, it presents a concept about flagship species. It is a particular species to represent the theatened habitat to the general public.
In the lecture, the professor uses a example to illustrate this the bird macaw.
When the trees were cut down and cleared away, the population of the bird declined and so did the other animals in the habitat.
So people strated spreading the word macow about how the bird needed help.
in the end, a lot of people contributed money and helpd the groups set up protected land.The bird and other animals were saved.
So this example illustrate the concept flagship species.
habitat
it is worth doing sth.
it is worth the money.
TPO24 6
In the lecture, the professor presents two kinds of narrators to illustrate how to tell a story.
the first narrator is objective narrator. from this kind , you can know what they do and what they say but that is all.
for example a man and a woman are going for a trip and from objective narrator you could only know what they say and what they do.You need to make readers to fill in the blank.
But in the second narraor,omniscient narrator you can know not only what they say and do, but also what they are thinking about.
For example , if the couple want to visit a friend and the man is thinking you can know the man is nervous because he hasn't seen his friend for a long time.
題目:Do you agree or disagree with the following statement:
The friends that you can have fun with are more important than the friends that you can get help from.
In my perspective, I think people who can help you are more important to someone's life because this kind of friend will play a key role to rescue you from bad situation you meet in life. And also, as you can image, in fact having fun is just one special kind of helping so according to those reasons I have to disagree the statement.
First of all, the friend who can help you will have a positive influence on you so that you can improve yourself. For example, when you are in university, a friend who can help you may study with you and you could discuss how to solve the assignment your teacher gives; And also, when you find a job, this kind of friend could teach you the skills you need in the company and this skill may be very significant so that you can have a good performance. However, another kind of friend can't do this, and in fact, usually they could have a negative effect on you because all things they know are having fun.
More importantly, a friend who can help you may give you a hand in emergency situation. In this situation, the most important thing is not your happiness but your life. All external expectation, all fear of embarrassment will fall away in the face of death. Supposing you are suffering a car accident, this kind of friend could call 911 at the first time and sent you to the hospital. But on the other hand, the other kind of friend may just watch how you are dying because they don't know how to do in this case. They know nothing but having fun. If you can't make sure your safety, according to Maslow's hierarchy of needs, you have no chance to pursue higher achievement.
But this kind of friend is still indispensable because they inevitably play a role in your life, you can receive your happiness when you get boring in your life. You can go shopping with him; you can watch movie with him; and also when you feel lonely you can go for a walk with him. But we need to emphasize is though this kind is also important, a friend who can help you is more significant in your life and you must treasure your relationship because it may last life long.
suffer
Rescue
Receive’
Negative positive
Indispensable :in/dis/pen/sable
Inevitably 不可避免的
Know - -
Emergency situation
Emphasize
Thief 小偷
1.題目:Do you agree or disagree with the following statement? People are now easier to become educated than in the past.
At present, a growing number of people will choose to study in university or at least in highschool so they could gain enough knowledge to live a good life. According to a famous survey conducted by a prestigious company called SurveyMonkey, %89.7 of people in the world have education experience in the college or in the highschool while the rate will decrease to %37.6 before World War 2.So in my opinion, I strongly agree the statement, people are easier to be educated now.
First of all, as you can image, with the development of the society, the people in 21st century have no worry about the physiological needs. It means we don't need to care much about food , water or heat source ,which is the most basic need for life and we can pay more attention on education or something you really have interest. According to Maslow's hierarchy of needs, that is to say ,it will be easy for you to pursue your self-actualization, like education. However, people in the past is easier to suffer war or disease. For example, supposing you are a boy living in World War 2, would you pay much attention on education when the food or shelter is still a big problem?
More importantly, the scientific power will play a significant role in education, making it easier for people to get education. For example, you can simply search materials you need on the internet by clicking your mouse several times; you can use MSN to communicate with professors all over the world without flying to the city where the professor are; and also you are able to use some multi-media devices to help you understand the complex questions such as projector. With the help of modern science and technology, there is no doubt that we can obtain a good education in a easy way.
However, we can't hold an absolute view about this issue because human itself also play a very important role in education. Even though there are a good environment and state-of-the-art technology in modern society, it will mean nothing if you just don't want to learn. But, over all, it is true that people are easier to be educated than in the past.
錯(cuò)誤單詞:
an absolute
Scientific
Simply
True
Disease
Multi-media
Materials
21st century 22nd,23rd
如果有哪位托福大神路過能幫我改下作文,感激不盡,必有重謝!^^