Correcting Face Distortion in Wide-Angle Videos

  title={Correcting Face Distortion in Wide-Angle Videos},
  author={Wei-Sheng Lai and Yichang Shih and Chia-Kai Liang and Ming-Hsuan Yang},
  journal={IEEE Transactions on Image Processing},
Video blogs and selfies are popular social media formats, which are often captured by wide-angle cameras to show human subjects and expanded background. Unfortunately, due to perspective projection, faces near corners and edges exhibit apparent distortions that stretch and squish the facial features, resulting in poor video quality. In this work, we present a video warping algorithm to correct these distortions. Our key idea is to apply stereographic projection locally on the facial regions. We… 



Distortion-free wide-angle portraits on camera phones

A new algorithm to undistort faces without affecting other parts of the photo is introduced, which locally adapts to the stereographic projection on facial regions, and seamlessly evolves to the perspective projection over the background.

Deep face normalization

Key applications of this deep learning framework range from robust image-based 3D avatar creation, portrait manipulation, to facial enhancement and reconstruction tasks for crime investigation, and it is found that the normalization results can be hardly distinguished from ground truth ones if the person is not familiar.

Fisheye Video Correction

This work presents an efficient and robust scheme for fisheye video correction, which minimizes time-varying distortion and preserves salient content in a coherent manner and illustrated with a range of examples, demonstrating coherent natural-looking video output.

Learning Perspective Undistortion of Portraits

The first deep learning based approach to remove perspective distortion artifacts from unconstrained portraits is presented, significantly improving the accuracy of both face recognition and 3D reconstruction and enables a novel camera calibration technique from a single portrait.

Motion-based video retargeting with optimized crop-and-warp

This work introduces a video retargeting method that achieves high-quality resizing to arbitrary aspect ratios for complex videos containing diverse camera and dynamic motions, and combines novel cropping and warping operators to find the best balance between the two operations.

Video stitching with spatial-temporal content-preserving warping

  • Wei JiangJinwei Gu
  • Computer Science
    2015 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
  • 2015
A novel algorithm for stitching multiple synchronized video streams into a single panoramic video with spatial-temporal content-preserving warping, which locally aligns frames from different videos while maintaining the temporal consistency is proposed.

Content-Aware Video Retargeting Using Object-Preserving Warping

Qualitative and quantitative analyses, including a user study and experiments on complex videos containing diverse cameras and dynamic motions, show a clear superiority of the novel content-aware warping approach over related video retargeting methods.

Scalable and coherent video resizing with per-frame optimization

This work proposes a new method that solves the scalability problem without compromising the resizing quality and matches the quality of state-of-the-art retargeting methods while dramatically reducing the computation time and memory consumption, making content-aware video resizing scalable and practical.

Content-preserving warps for 3D video stabilization

A technique that transforms a video from a hand-held video camera so that it appears as if it were taken with a directed camera motion, and develops algorithms that can effectively recreate dynamic scenes from a single source video.

Dynamic Video Stitching via Shakiness Removing

This paper proposes a novel approach of video stitching and stabilization for videos captured by mobile devices, and proposes a method to distinguish between right and false matches, and encapsulate the false match elimination scheme and the optimization into a loop to prevent the optimization from being affected by bad feature matches.