1. 제    목 : Multi-Scale Generalized PlaneMatch based occlusion detection and correspondence for optical flow

2. 발 표 자 : Inchul Choi

3. 일    시 : 2018년  11월  22일(목)  16:00 ~ 17:30

4. 장    소 : 경북대학교 IT대학 1호관 318호

5. 초청교수 : 이민호 교수

6. 강사약력 :

  · Ph.D in Computer Science 2018 Concentrations : Machine Learning , Computer Vision,
    Data  Science Dissertation : Multi-Scale Generalized Plane Match for Optical Flow 
  · Master of Science from CISE (Computer & Information Science and Engineering)Department,
    University of Florida, Aug, 2009~ 2011 
  · Master of Science, Computer Engineering, Pohang University of Science and Technology
    (Postech) at Pohang, Korea, 2003 Concentrations : System Software Master Thesis :
    Enhancing Prediction Accuracy in PCM-based File Prefetch by  Constrained Pattern Replacement Algorithm 
  · Bachelor of Science in Engineering, Electronic and Electrical Engineering,
    Kyungpook National University at Daegu, Korea,2001

7. 내용요약 :

  Despite recent advances, estimating optical flow remains a challenging problem in the presence of illumination change, large occlusions or fast movement. In this paper, we propose a novel optical flow estimation framework which can provide accurate dense correspondence and occlusion localization through a multi-scale generalized plane matching approach. In our method, we regard the scene as a collection of planes at multiple scales, and for each such plane, compensate motion in consensus to improve match quality. We estimate the square patch plane distortion using a robust plane model detection method and iteratively apply a plane matching scheme within a multi-scale framework. During the flow estimation process, our enhanced plane matching method also clearly localizes the occluded regions. In experiments on MPI-Sintel datasets, our method robustly estimated optical flow from given noisy correspondences, and also revealed the occluded regions accurately. Compared to other state-of-the-art optical ow methods, our method shows accurate occlusion localization, comparable optical ow quality, and better thin object detection.

◀ 문의처 : IT대학 전자공학부 인공두뇌연구실 이혜경 ☎ 940-8616