Survey of Background Blurring Shadow 2008.10.07

Outline Motivation Introduction Related work/Paper survey System design Todo list

Motivation Image enhancement 想要凸顯主題， 但是DSLR很貴、鏡頭很貴…… 景的閒雜人物很討厭 For privacy preserving 利用refocus來達到image轉成movie的效果

Introduction Depth of field （景深） 焦距 光圈 物距 Bokeh (散景)

By DSLR

By DC

By Photoshop

Related Work Refocusing : Active Refocusing of Images and Videos [siggraph07]

Related Work Refocusing : Active Refocusing of Images and Videos [siggraph07] Depth hint: active project light Blur kernel: Gaussian blur

Related Work Image and Depth from a Conventional Camera with a Coded Aperture [siggraph07]

Coded aperture- reduce uncertainty in scale identification Related Work Coded aperture- reduce uncertainty in scale identification Conventional Coded Larger scale Correct scale So why a coded aperture does actually helps us to estimate depth? We captured the same image with both a conventional and a coded lens. And let’s try again to deconvolve our local image window with different scales of the defocus filter. With a conventional aperture we have uncertainty between the correct scale and the smaller one, but with the coded aperture, both the smaller and the larger scales are making mess, so the uncertainty is reduced. This uncertainty demonstrates the major idea behind the coded aperture, and for those of you who want to understand better what happens, we make a short transition into the frequency domain. Smaller scale

Using the all-focused image and the layered depth map, one application is post-exposure refocusing.

Related Work Single Image Dehazing [siggraph08] Radiative transport equation 照片裡有霧 光線透過霧會產生分散的效果 Addtive term  加上去的效果 一開始的model是J 是表示原先真實場景的光線，經由t這個transmit的function會產生某種程度上的分散，再加上airlight霧的光線顏色vector 將這個model簡化成有同樣反照率的表層，他們有同樣的顏色，但是有不同的shade程度，所以J(x) => l(x)R

Related Work Single Image Dehazing [siggraph08] 照片裡有霧 光線透過霧會產生分散的效果 Addtive term  加上去的效果 一開始的model是J 是表示原先真實場景的光線，經由t這個transmit的function會產生某種程度上的分散，再加上airlight霧的光線顏色vector 將這個model簡化成有同樣反照率的表層，他們有同樣的顏色，但是有不同的shade程度，所以J(x) => l(x)R

Related Work Single Image Dehazing [siggraph08]

Related Work 要做refocusing  要找出depth map Usage of Depth map Reconstruct 3D scene 3D shape recovery

Related Work Depth Recovery Depth from stereo Structure from motion Shape from shading Range from focus Depth from defocus

Related Work How to find depth map Perspective of two images Blur effect of an image Other hint Do we really need a exact depth map?

System design

System design Object segmentation Image matting : Hard cut Soft cut Grab cut Lazy snapping Watershed Mean-shift Soft cut Tri-map alpha matting Color segmentation

System design Depth map reconstruction Vanish point or horizontal line

System design Refocus the Image different blur kernel

Todo List 有depth map  產生出有淺景深效果的result 有object segment, subject 等hint 產生depth map Object segmentation (with/without user hint)

Application