Iterative Optimization

Slides:



Advertisements
Similar presentations
Bayesian Belief Propagation
Advertisements

The fundamental matrix F
CSCE643: Computer Vision Bayesian Tracking & Particle Filtering Jinxiang Chai Some slides from Stephen Roth.
Exploration of bump, parallax, relief and displacement mapping
SOFT SCISSORS: AN INTERACTIVE TOOL FOR REALTIME HIGH QUALITY MATTING International Conference on Computer Graphics and Interactive Techniques ACM SIGGRAPH.
I Images as graphs Fully-connected graph – node for every pixel – link between every pair of pixels, p,q – similarity w ij for each link j w ij c Source:
1.  Texturing is a core process for modeling surface details in computer graphics applications › Texture mapping › Surface texture synthesis › Procedural.
Activity Recognition Aneeq Zia. Agenda What is activity recognition Typical methods used for action recognition “Evaluation of local spatio-temporal features.
Real-Time Human Pose Recognition in Parts from Single Depth Images Presented by: Mohammad A. Gowayyed.
Face Alignment at 3000 FPS via Regressing Local Binary Features
Video Matting from Depth Maps Jonathan Finger Oliver Wang University of California, Santa Cruz {jfinger,
GrabCut Interactive Image (and Stereo) Segmentation Carsten Rother Vladimir Kolmogorov Andrew Blake Antonio Criminisi Geoffrey Cross [based on Siggraph.
A Gimp Plugin that uses “GrabCut” to perform image segmentation
Robust Foreground Detection in Video Using Pixel Layers Kedar A. Patwardhan, Guillermoo Sapire, and Vassilios Morellas IEEE TRANSACTION ON PATTERN ANAYLSIS.
Jue Wang Michael F. Cohen IEEE CVPR Outline 1. Introduction 2. Failure Modes For Previous Approaches 3. Robust Matting 3.1 Optimized Color Sampling.
Stephen J. Guy 1. Photomontage Photomontage GrabCut – Interactive Foreground Extraction 1.
GrabCut Interactive Image (and Stereo) Segmentation Joon Jae Lee Keimyung University Welcome. I will present Grabcut – an Interactive tool for foreground.
3D Video Generation and Service Based on a TOF Depth Sensor in MPEG-4 Multimedia Framework IEEE Consumer Electronics Sung-Yeol Kim Ji-Ho Cho Andres Koschan.
Learning to Detect A Salient Object Reporter: 鄭綱 (3/2)
HCI Final Project Robust Real Time Face Detection Paul Viola, Michael Jones, Robust Real-Time Face Detetion, International Journal of Computer Vision,
1 Image Completion using Global Optimization Presented by Tingfan Wu.
1 Abstract This paper presents a novel modification to the classical Competitive Learning (CL) by adding a dynamic branching mechanism to neural networks.
CSCE 641 Computer Graphics: Image-based Modeling Jinxiang Chai.
Probabilistic video stabilization using Kalman filtering and mosaicking.
Natural Video Matting with Depth Jonathan Finger Oliver Wang University of California, Santa Cruz {jfinger,
Adapted from: CULLIDE: Interactive Collision Detection Between Complex Models in Large Environments using Graphics Hardware Naga K. Govindaraju, Stephane.
A Closed Form Solution to Natural Image Matting
High-Quality Video View Interpolation
Abstract Extracting a matte by previous approaches require the input image to be pre-segmented into three regions (trimap). This pre-segmentation based.
UNC Chapel Hill M. C. Lin Overview of Last Lecture About Final Course Project –presentation, demo, write-up More geometric data structures –Binary Space.
An Iterative Optimization Approach for Unified Image Segmentation and Matting Hello everyone, my name is Jue Wang, I’m glad to be here to present our paper.
IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 20, NO. 11, NOVEMBER 2011 Qian Zhang, King Ngi Ngan Department of Electronic Engineering, the Chinese university.
Texture Optimization for Example-based Synthesis
REU Project 4D Efficient Real-Time Graph Based Temporal Point Cloud Segmentation Steven Hickson.
Combined Central and Subspace Clustering for Computer Vision Applications Le Lu 1 René Vidal 2 1 Computer Science Department, Johns Hopkins University,
Scientific Writing Abstract Writing. Why ? Most important part of the paper Number of Readers ! Make people read your work. Sell your work. Make your.
Global Parametrization of Range Image Sets Nico Pietroni, Marco Tarini, Olga Sorkine, Denis Zorin.
Human pose recognition from depth image MS Research Cambridge.
Efficient Visual Object Tracking with Online Nearest Neighbor Classifier Many slides adapt from Steve Gu.
1 Markov random field: A brief introduction (2) Tzu-Cheng Jen Institute of Electronics, NCTU
Graphcut Textures Image and Video Synthesis Using Graph Cuts
Motion Estimation using Markov Random Fields Hrvoje Bogunović Image Processing Group Faculty of Electrical Engineering and Computing University of Zagreb.
Efficient Belief Propagation for Image Restoration Qi Zhao Mar.22,2006.
Video Textures Arno Schödl Richard Szeliski David Salesin Irfan Essa Microsoft Research, Georgia Tech.
CSCI 631 – Foundations of Computer Vision March 15, 2016 Ashwini Imran Image Stitching.
Matte-Based Restoration of Vintage Video 指導老師 : 張元翔 主講人員 : 鄭功運.
Processing Images and Video for An Impressionist Effect Automatic production of “painterly” animations from video clips. Extending existing algorithms.
CSCI 631 – Foundations of Computer Vision March 15, 2016 Ashwini Imran Image Stitching Link: singhashwini.mesinghashwini.me.
A Globally Optimal Algorithm for Robust TV-L 1 Range Image Integration Christopher Zach VRVis Research Center Thomas Pock, Horst Bischof.
Physically-Based Motion Synthesis in Computer Graphics
Optimal Acceleration and Braking Sequences for Vehicles in the Presence of Moving Obstacles Jeff Johnson, Kris Hauser School of Informatics and Computing.
Imaging and Depth Estimation in an Optimization Framework
Interactive Offline Tracking for Color Objects
Jun Shimamura, Naokazu Yokoya, Haruo Takemura and Kazumasa Yamazawa
GrabCut Interactive Foreground Extraction using Iterated Graph Cuts Carsten Rother Vladimir Kolmogorov Andrew Blake Microsoft Research Cambridge-UK.
Recognizing Deformable Shapes
Computer Vision, Robotics, Machine Learning and Control Lab
Introduction to Graphics Modeling
Robust Visual Motion Analysis: Piecewise-Smooth Optical Flow
Fast Preprocessing for Robust Face Sketch Synthesis
Using Flow Textures to Visualize Unsteady Vector Fields
School of Electrical and
Xiang Liang Natural Image Matting Xiang Liang
PRAKASH CHOCKALINGAM, NALIN PRADEEP, AND STAN BIRCHFIELD
Online Graph-Based Tracking
Biointelligence Laboratory, Seoul National University
WELCOME.
Problem Image and Volume Segmentation:
Random Neural Network Texture Model
Initial Progress Report
Presentation transcript:

Iterative Optimization Easy Matting Model the unknown region as a Markov Random Field. Introduce a local refinement technique to manipulate the continuous energy field in selected local regions. Energy-driven scheme can be extended to video matting. Specifically, we propose an energy minimization framework for interactive image matting. As shown in this example, we start from a few strokes, and iteratively estimate the matte until it converges. Notice the fact that no explicit trimap is required. Beginning with the known region, we model the unknown region as a Markov Random Field (MRF) and formulate its energy as the combination of one data term and one smoothness term. Second, we introduce a novel local refinement technique to manipulate the continuous energy field in selected local regions. The modified local regions can be seamlessly integrated into the final result. Lastly, our approach can be directly extended to video matting by considering an additional temporal smoothness term, with which the spatio-temporal smoothness is faithfully preserved. Iterative Optimization Initial Input Final Matte

Results Knockout 2 Poisson Bayesian Input image Trimap Strokes Now I will show some results of our system. We can use bayesian, Knockout 2 and Poisson matting to extract matte from this image based on the same trimap. However, they yield more visual artifacts than our approach, even with more user inputs. Starting from a few strokes, BP matting didn’t give us the right matte, however, our system can generate a good matte. Strokes BP Matting Global Easy Matting

Conservative Voxelization Conservative correctness: all voxels intersecting the input model are recognized. Efficient and robust implementation in the GPU. No preprocessing required. Previous approach: generate a single voxel for each pixel by using the depth in the pixel center Our approach: generate multiple voxels for each pixel by computing the depth range in the pixel

Application to Collision Detection Efficient (in real-time) Support deformable models Conservative correctness: colliding voxels refer to potentially colliding regions non-colliding voxels refer to regions with no intersection Collision detection between the buddha model (210k triangles) and the morphing hand model (5k triangles) is accomplished in 114 ms (~8.8 fps)

Data-driven Tree Animation Synthesis Adapt the motion synthesis algorithm in Human animation to tree animation. Advantages: realistic & efficient Contributions: A practical sampling algorithm leading to a rich and reusable motion database; Improved algorithm for motion graph construction; Efficient algorithm for motion synthesis which has a fast response to user interaction.

Dynamic Forest Scene Demo