Method to improve quality of static scene in video using photographs

1
Method to improve quality of static scene in
video using photographs

• Hari Krishna

2
Digital Photography

• High Dynamic Range (HDR)
• Better noise reduction and speed
• High resolution

3
Key measures for video

• Frame rate
• Image Longevity
• Color reproduction

4
Procedure..

• With the right tools the user can transfer
  desirable qualities from the photographs to the
  video
• Scene motion, lighting and view point
  characteristics can drastically change the
  appearance of the scene between photographs and
  videos
• We take high quality photographs to form a
  static scene of the video by using right
  authoring tool and an algorithm which establishes
  the correspondence between the video and
  photographs

5
This system consists of..

• Object removal of unwanted scenes
• This method stabilizes the video by smoothing
  and re-render the scene by using Poisson Image
  editing and iterative optimization algorithm
• Camera Shake
• Structure from motion data is computed. Video
  edits are done followed by semantic matching and
  local context matching
• Color harmonization
• Tools for color harmonization are provided.
• Noise, poor exposure and tone mapping can be
  reduced

6

• Combining videos and photographs is the main
  idea of the framework
• The technique for automatically combining
  photographs and videos of a static scene creates
  many interesting effects
• In this framework I restrict to the scenes
  with static geometry

7
Framework uses..

• Multi-view stereo algorithm (MVR)
• Technical improvements are made in order to
  compute view-dependent depths and establish pixel
  to pixel correspondence using structure from
  motion data.
• Image based rendering (IMR)
• This approach uses graph cuts and gradient
  domain compositing while taking camera motion
  into account and re-renders the input video and
  then temporal and spatial dynamics of the
  photographs

8
Framework uses.. (contd.)

• Scene Completion Algorithm (SCR)
• Restructuring the images in a video frame and
  optimization of images to generate final high
  quality output video.

9
System Flow..
10
Any Questions ?
Thank You !!