Download presentation
Presentation is loading. Please wait.
Published byAmbrose Phillips Modified over 6 years ago
1
Computer Vision, Robotics, Machine Learning and Control Lab
René Vidal Center for Imaging Science, Whitaker Institute of Biomedical Engineering, Johns Hopkins University Add d15.avi
2
Static scene reconstruction
Image Segmentation Penguin/Ice/Water Face Recognition Man/Women Video Segmentation Tiger/Water/Bushes 3-D Reconstruction One of the reasons we are interested in GPCA is because there are various problems in computer vision that have to do with the simultaneous estimation of multiple models from visual data. Consider for example segmenting an image into different regions based on intensity, texture of motion information. Consider also the recognition of various static and dynamic processes such as human faces or human gaits from visual data. Although in this talk I will only consider the first class of problems, it turns out that, at least from a mathematical perspective, all the above problems can be converted into following generalization of principal component analysis, which we conveniently refer to as GPCA
3
Dynamic scene reconstruction
Temporal segmentation Guest/Host/Both Dynamic texture recognit Water/Steam/Foliage/Fire Spatial segmentation Car/Background Gait/Posture recognition Running/Walking/Limping One of the reasons we are interested in GPCA is because there are various problems in computer vision that have to do with the simultaneous estimation of multiple models from visual data. Consider for example segmenting an image into different regions based on intensity, texture of motion information. Consider also the recognition of various static and dynamic processes such as human faces or human gaits from visual data. Although in this talk I will only consider the first class of problems, it turns out that, at least from a mathematical perspective, all the above problems can be converted into following generalization of principal component analysis, which we conveniently refer to as GPCA
4
Why are these problems challenging?
“Chicken-and-egg” problems Given segmentation, estimate models Given models, segment the data Easy for humans, difficult for a computer Given a set of data points lying on a collection of linear subspaces, without knowing which point corresponds to which subspace and without even knowing the number of subspaces, estimate the number of subspaces, a basis for each subspace and the segmentation of the data. As stated, the GPCA problem is one of those challenging CHICKEN-AND-EGG problems in the following sense. If we knew the segmentation of the data points, and hence the number of subspaces, then we could solve the problem by applying standard PCA to each group. On the other hand, if we had a basis for each subspace, we could easily assign each point to each cluster. The main challenge is that we neither know the segmentation of the data points nor a basis for each subspace, and the questions is how to estimate everything from the data only. Previous geometric approaches to GPCA have been proposed in the context of motion segmentation, and under the assumption that the subspaces do NOT intersect. Such approaches are based on first clustering the data, using for example K-means or spectral clustering techniques, and then applying standard PCA to each group. Statistical methods, on the other hand, model the data as a mixture model whose parameters are the mixing proportions and the basis for each subspace. The estimation of the mixture is a (usually non-convex) optimization problem that can be solved with various techniques. One of the most popular ones is the expectation maximization algorithm (EM) which alternates between the clustering of the data and the estimation of the subspaces. Unfortunately, the convergence of such iterative algorithms depends on good initialization. At present, initialization is mostly done either at random, using another iterative algorithm such as K-means, or else in an ad-hoc fashion depending on the particular application. It is here where our main motivation comes into the picture. We would like to know if we can find a way of initializing iterative algorithms in an algebraic fashion. For example, if you look at the two lines in this picture, there has to be a way of estimating those two lines directly from data in a one shot solution, at least in the ABSENCE of noise.
5
What is our approach? x = A v y C w b = D p ( x ) j
Representation: hybrid dynamical models Computational solution: Generalized Principal Component Analysis (GPCA) algebraic geometry, differential geometry, projective geometry and linear algebra dynamics x t + 1 = A v y C w images appearance running walking Theorem: GPCA [Vidal et al. 2004] b i = D p n ( x ) j z
6
Texture-based image segmentation
Our approach Human
7
Texture-based image segmentation
8
Motion-based video segmentation
9
Motion-based video segmentation
Sequence A Sequence B Sequence C Percentage of correct classification
10
Temporal video segmentation
Segmenting N=30 frames of a sequence containing n=3 scenes Host Guest Both Image intensities are output of linear system Apply GPCA to fit n=3 observability subspaces dynamics appearance images x t + 1 = A v y C w
11
Temporal video segmentation
Segmenting N=60 frames of a sequence containing n=3 scenes Burning wheel Burnt car with people Burning car Image intensities are output of linear system Apply GPCA to fit n=3 observability subspaces dynamics appearance images x t + 1 = A v y C w
12
Segmentation of dynamic textures
Output of a linear dynamical model (Soatto et al. 2001) Dynamic texture segmentation using level-sets (Doretto et al. 2003) dynamics x t + 1 = A v y C w images appearance
13
Segmentation of dynamic textures
One dynamic texture lives in the observability subspace Multiple textures live in multiple subspaces Apply PCA to image intensities Apply GPCA to projected data images
14
Segmentation of moving dynamic textures
Time varying model Segmentation of multiple moving subspaces Apply PCA to intensities in a moving time window Apply GPCA to projected data dynamics images appearance
15
Segmentation of moving dynamic textures
Static textures: optical flow is computed from brightness constancy constraint (BCC) Dynamic textures: optical flow computed from dynamic texture constancy constraint (DTCC)
16
MRI-based heart motion analysis
Segmentation: extract static and dynamic landmarks from high-resolution images Static (chest wall, aorta, spinal cord, sternum) Dynamic (heart) Registration: match real-time low-resolution images to high-quality images Rigid (chest wall, aorta, spinal cord, sternum) Non-rigid (heart) Temporal alignment (using cardiac cycle) Dynamical To synthesize MRI data with ECG data ECG gating
17
Dynamical Chest and Heart Segmentation
Original Image Morphological Operations Dynamic Segmentation Rene Segmentation By Intensity
18
Heart Segmentation by Dynamic Textures
nash
19
Chest Segmentation by Dynamic Textures
nash
20
DTI-based spinal cord injury detection
Diffusion tensor imaging: at each voxel have Intensity of the voxel Orientation of the voxel: tensor/ellipsoid Can reconstruct fibers in the tissue Spinal cord injury detection: regions where fibers no longer exist
21
DTI-based spinal cord injury detection
Axial segmentation of fibers in white matter Injury detection
22
Ongoing and future research
Computer Vision & Graphics Multiview motion segmentation Cue integration (motion and texture) for segmentation Shape recognition (faces) Recognition of human motion and human activities Animation of articulated bodies Synthesis of dynamic textures Biomedical Applications Heart motion analysis Spinal cord injury detection Locomotion, posture and gait analysis Medical robotics: ERC-CISST Center for Computer Integrated Surgical Systems and Technology Vision, Learning and Statistical Geometry Robust GPCA Model selection Relations to Kernel PCA Estimating mixtures of dynamical models Estimating manifolds from sample data points Biometrics: Perception Human Crowd Activities
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.