2017年09月_seamanj

12月 11月 10月 09月 08月 06月 04月 03月 02月 01月

原创 so(3) se(3) optimization

http://www.seas.upenn.edu/~cjtaylor/PUBLICATIONS/pdfs/TaylorTR94b.pdf

2017-09-28 17:19:06 2368

原创 Modeling Deformable Objects from a Single Depth Camera

1 partial surface →(point correspondence)→ warp partial surface →(assemble)→complete watertight surface2 section 3In the initialization step, … into 3D meshes.颜色部分用SIFT提取salient features 深度部分用来进行3D

2017-09-19 04:08:40 400

原创 convex optimization from stanford

视频来源: https://www.youtube.com/watch?v=XFKBNJ14UmY 课程来源: http://www.stat.cmu.edu/~ryantibs/convexopt/Lecture 1 Introduction 1) 15:20 least squares is easier to solve than least absolute deviations,

2017-09-14 18:24:57 656

原创 Sparse Modeling of Intrinsic Correspondences

这篇文章的主要贡献是把区域的匹配细化到点的匹配上去As the input to our algorithm, we are given only two sets of regions in two shapes; no descriptors are provided so the correspondence between the regions is not know, nor w

2017-09-14 01:03:38 282

原创 coupled quasi-harmonic bases

1) section 2, 公式WΦ=DΦΛ" role="presentation" style="position: relative;">WΦ=DΦΛWΦ=DΦΛW\Phi = D\Phi\Lambda推导LΦ=ΦΛ" role="presentation" style="position: relati

2017-09-13 01:53:17 404

原创 map-based exploration of intrinsic shape differences and variability

这篇主要用矩阵来测量两个shape之间的area distortion和comformal distortion2) 图二中V(f)V(f)说明how much ff is distorted by FF, 3)Theorem 2中, f,gf,g都作用于MM上 4)公式(3)的推导根据Theorem 2, fTHMDg=fTFTHNFgf^TH^MDg = f^TF^TH^NFg HM

2017-09-09 17:49:04 492

原创 The Wave kernel Signature: A Quantum Mechanical Approach to shape Analysis

相对于HKS, WKS是模拟粒子在某一时刻tt,出现在地点为xx的概率,表示为ψ(x,t)\psi(x,t) HKS的缺点是它强调了低频部分, 而忽略了高频的影响. 为了解决这个问题WKS在方程里面加了一个ii, 变成复数. {引用[1],66页 A mathematical solution is to multiply the Laplace-Beltrami operator by the

2017-09-06 21:12:47 1373 3

原创 Function Maps: A Flexible Representation of Maps Between Shapes

1) 图1，先通过discriptor算出每点的值，然后算出function map，即matrix C" role="presentation" style="position: relative;">CCC. 左图是说明discriptor每有区分左右，中图加了对应约束使其左右交换，右图为了说明function map满足代数运算，左图减去中图就得到保方向的function map

2017-09-05 21:48:20 1188

原创 A Concise and Provably Informative Multi-Scale Signature Based on Heat Diffusion

这篇文章有很多理论基础我没有看懂, 感兴趣的同学请参考HSU02这本书1) 先说figure 1, 曲线图是描述不同时间段, scaled HKS的值, 最开始1,2,3,4这四点相同, 因为这时还没有扩散, 到最后1,2 变成一类, 3,4 变成一类, 这是因为,1,2扩散到了头, 而3,4,扩散到了尾, 而头和尾有很大区别, 所以出现情况描述的情况2) 第3页 Thus bo

2017-09-02 04:04:19 877