Source author record

Yichao Xu

Yichao Xu appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Computer Vision

Catalog footprint

What is connected

2works

1topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2015arXiv

TransCut: Transparent Object Segmentation from a Light-Field Image

The segmentation of transparent objects can be very useful in computer vision applications. However, because they borrow texture from their background and have a similar appearance to their surroundings, transparent objects are not handled well by regular image segmentation methods. We propose a method that overcomes these problems using the consistency and distortion properties of a light-field image. Graph-cut optimization is applied for the pixel labeling problem. The light-field linearity is used to estimate the likelihood of a pixel belonging to the transparent object or Lambertian background, and the occlusion detector is used to find the occlusion boundary. We acquire a light field dataset for the transparent object, and use this dataset to evaluate our method. The results demonstrate that the proposed method successfully segments transparent objects from the background.

preprint2014arXiv

Mobile Camera Array Calibration for Light Field Acquisition

The light field camera is useful for computer graphics and vision applications. Calibration is an essential step for these applications. After calibration, we can rectify the captured image by using the calibrated camera parameters. However, the large camera array calibration method, which assumes that all cameras are on the same plane, ignores the orientation and intrinsic parameters. The multi-camera calibration technique usually assumes that the working volume and viewpoints are fixed. In this paper, we describe a calibration algorithm suitable for a mobile camera array based light field acquisition system. The algorithm performs in Zhang's style by moving a checkerboard, and computes the initial parameters in closed form. Global optimization is then applied to refine all the parameters simultaneously. Our implementation is rather flexible in that users can assign the number of viewpoints and refinement of intrinsic parameters is optional. Experiments on both simulated data and real data acquired by a commercial product show that our method yields good results. Digital refocusing application shows the calibrated light field can well focus to the target object we desired.