Computational Photography and Videography
Computational photography is an emerging field, aiming to use computational techniques to break a camera's limitations on its resolution, aspect ratio, dynamic range, field of view and quality. Along this line, we had the following research results.
We have also investigated methods to animate still
pictures into video textures. The results was published in SIGGRAPH 2005.
High dynamic range imaging. Recovering an HDR image from differently exposed photographs has become very popular. However, it often requires a tripod to keep the camera still because, when taking photos with a hand-held camera, long-exposed photographs often become blurred. We proposed a method to allow HDR photo taking with hand-held cameras by iteratively performing blur kernel estimation, HDR image reconstruction and camera curve recovery. When convergence, we simultaneously obtain an HDR image with rich and clear structures, the camera response curve and blur kernels. The proposed algorithm makes easy HDR image taking. It was accepted by CVPR 2009.
Casual videos taken by amateurs often exhibits annoying jitter due to shaky motion of an unsteady handheld camera.
Video stabilization removes unwanted video perturbations due to unstable camera motions for enhancing user's experiences in watching casual videos.
In ICCV 2009, we proposed an approach to stabilize a video by smoothing feature trajectories without explicitly estimating camera motion.
The method is more robust than previous methods which often rely on motion estimation.
Image Resizing. Image resizing has received considerable attention recently and many content-aware methods have been proposed. However, few specifically pay attention to preserve line structure properties. Since human are often very sensitive to distortions of geometric structures such as lines, such distortions often look more noticeable and disturbing. We proposed an image resizing method to simultaneously minimize content distortion of prominent regions and preserve important line-structure properties: parallelism, collinearity and orientation. It has been accepted by CVPR 2012
Moire Pattern Synthesis. In ICCV 2013, we proposed a method for target-driven moire pattern synthesis:
Given a target image I, find two curvilinear grating images L1 and L2 such that the moire pattern of their superposition is close to I.
The technique can be used for generating moire arts and moire cryptography.
Shape-preserving warps. In CVPR 2014, we proposed a novel parametric warp which is a spatial combination of a projective transformation and a similarity transformation.
The proposed warp has the strengths of both projective and similarity warps.
It provides good alignment accuracy as projective warps while preserving the perspective of individual image as similarity warps.
It can also be combined with more advanced local-warp-based alignment methods such as the as-projective-as-possible warp for better alignment accuracy.
With the proposed warp, the field of view can be extended by stitching images with less projective distortion (stretched shapes and enlarged sizes).
Image deblurring. In CVPR 2015, we propose a single-image blur kernel estimation algorithm that utilizes the normalized color-line prior to restore sharp edges without altering edge structures or enhancing noise. A comprehensive evaluation on a large image deblurring dataset shows that our algorithm achieves the state-of-the-art results.
- Blur Kernel Estimation using Normalized Color-Line Priors
- CVPR 2015
- Shape-Preserving Half-Projective Warps for Image Stitching
- CVPR 2014
- Target-Driven Moire Pattern Synthesis by Phase Modulation
- ICCV 2013
- Rectangling Stereographic Projection for Wide-Angle Image Visualization
- ICCV 2013
- A Line-Structure-Preserving Approach to Image Resizing
- CVPR 2012
Stabilization using Robust Feature Trajectories
- ICCV 2009
- High Dynamic
Range Image Reconstruction from Hand-held Cameras
- CVPR 2009
- Animating Pictures
with Stochastic Motion Textures
Dan B Goldman,
Ke Colin Zheng,
- SIGGRAPH 2005
This research is supported by:
cyy -a-t- csie.ntu.edu.tw