1 / 16

Layered Texture Matting

Layered Texture Matting. 9562631 葉展岱 9565524 范智勝. Outline. Introduction Implementation Overview Current Work Reference. Introduction. Our goal Doing texture synthesis on various objects with an irregular shape. Without losing the 3D property in vision.

nhi
Download Presentation

Layered Texture Matting

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Layered Texture Matting 9562631 葉展岱 9565524 范智勝

  2. Outline • Introduction • Implementation Overview • Current Work • Reference

  3. Introduction • Our goal • Doing texture synthesis on various objects with an irregular shape. • Without losing the 3D property in vision. • ( differ from texture transfer) • Preserving the property of texture by using the layered texture synthesis. • E.g., leaves, hair

  4. Implementation Overview • Shape from Shading - 葉展岱 • (Patch Orientation) • Texture Synthesis - 范智勝 • (Layered Texture Synthesis)

  5. Shape from Shading(1) • Normal Recovery • Related parameters • S:a unit vector toward a sufficiently distant point light source • I(x,y):intensity of pixel at (x,y). • Imin:minimum intensity • ambient light in the scene • Imax:maximum intensity • faces the light source

  6. Shape from Shading(2) • c(x,y) = (I(x,y)−Imin)/(Imax −Imin) • cosine of the angle of incidence at (x,y). • s(x,y) = sqrt(1−c(x,y)^2) • sine of the angle of incidence at (x,y). • G(x,y) = ▽I(x,y)−(▽I(x,y) ·S)S • ▽I(x,y) is the image gradient. • N(x,y) = c(x,y)*S+s(x,y)*G(x,y)/||G(x,y)|| So, what we need is ‘S’.

  7. Shape from Shading(3) • How to estimate S ? N(xi,yi).S = (I(xi,yi) – Imin)/(Imax - Imin) • (xi,yi): on the boundary of the object’s projection • N(xi,yi): the gradient of (xi,yi) • The source vector S is the least-squares solution to the over constrained linear system

  8. Shape from Shading(2) • Surface Segmentation • Cluster with the pixels with similar normals. • Use Watershed as cluster method.

  9. Texture Synthesis • Image Quilting Procedure • Define similarity using L2-norm applied to every pixel/color in block. • Optimize overlap region using minimum error boundary cut.

  10. Texture Synthesis(without cluster)

  11. Current Work(1) • Original Image & texture

  12. Current Work(2) • Cut out the target image • Use Matting

  13. Current Work(3) • Normal Recovery • Translate N(x,yz) to (R,G,B)

  14. Current Work(4) • Patch Cluster • Use Watershed as cluster method

  15. Current Work(5) • Texture Synthesis • Obey the normals and intensity

  16. Reference • [1] H. Fang and JC. Hart. Texture Synthesis as Photograph Editing SIGGRAPH 2004. • [2] A. A. Efros and WT. Freeman. Image Quilting for Texture Synthesis and Transfer. SIGGRAPH 2001. • [3] A. Levin, D. Lischinski, and Y. Weiss. A Closed Form Solution to Natural Image Matting. CVPR 2006.

More Related