We address the problem of deformable shape and motion recovery from point correspondences in multiple perspective images. We use the low-rank shape model, i.e. the 3D shape is represented as a linear combination of unknown shape bases.We propose a new way of looking at the low-rank shape model. Instead of considering it as a whole, we assume a coarse-to-fine ordering of the deformation modes, which can be seen as a model prior. This has several advantages. First, the high level of ambiguity of the original low-rank shape model is drastically reduced since the shape bases can not anymore be arbitrarily re-combined. Second, this allows us to propose a coarse-to-fine reconstruction algorithm which starts by computing the mean shape and iteratively adds deformation modes. It directly gives the sought after metric model, thereby avoiding the difficult upgrading step required by most of the other methods. Third, this makes it possible to automatically select the number of deformation modes as the reconstruction algorithm proceeds. We propose to incorporate two other priors, accounting for temporal and spatial smoothness, which are shown to improve the quality of the recovered model parameters.The proposed model and reconstruction algorithm are successfully demonstrated on several videos and are shown to outperform the previously proposed algorithms.

No abstract

No abstract

Abnormal event detection is an important issue in video surveillance applications. The goal is to detect abnormal or suspicious behaviors while given training samples that contain only normal behaviors. Sparse representation has showed its effectiveness in abnormal event detection [2,3,4,5], where a dictionary is commonly learned during training and anomalies are detected based on reconstruction error from the learned dictionary. Note that only a small proportion of the data is used for trainingrelatively to the huge amount of surveillance data, there is a high risk of incomplete normal patterns in the training data. Consequently, dictionary learning is crucial to the overall abnormality detection performance.We propose a Behavior-Specific Dictionary (BSD) algorithm, which takes into consideration the relation of atoms in one dictionary without 1,a1,2,...a1,d1,

No abstract

No abstract

Traditional forest restoration (FR) monitoring methods employ spreadsheets and photos taken at the ground level. Since remotely piloted aircraft (RPA) generate a panoramic high resolution and georeferenced view of the entire area of interest, this technology has high potential to improve the traditional FR monitoring methods. This study evaluates how low-cost RPA data may contribute to FR monitoring of the Brazilian Atlantic Forest by the automatic remote measurement of Tree Density, Tree Height, Vegetation Cover (area covered by trees), and Grass Infestation. The point cloud data was processed to map the Tree Density, Tree Height, and Vegetation Cover parameters. The orthomosaic was used for a Random Forest classification that considered trees and grasses as a single land cover class. The Grass Infestation parameter was mapped by the difference between this land cover class (which considered trees and grasses) and the Vegetation Cover results (obtained by the point cloud data processing). Tree Density, Vegetation Cover, and Grass Infestation parameters presented F_scores of 0.92, 0.85, and 0.64, respectively. Tree Height accuracy was indicated by the Error Percentage considering the traditional fieldwork and the RPA results. The Error Percentage was equal to 0.13 and was considered accurate because it estimated a 13% shorter height for trees that averaged 1.93 m tall. Thus, this study showed that the FR structural parameters were accurately measured by the low-cost RPA, a technology that contributes to FR monitoring. Despite accurately measuring the structural parameters, this study reinforced the challenge of measuring the Biodiversity parameter via remote sensing because the classification of tree species was not possible. After all, the Brazilian Atlantic Forest is a biodiversity hotspot, and thus different species have similar spectral responses in the visible spectrum and similar geometric forms. Therefore, until improved automatic classification methods become available for tree species, traditional fieldwork remains necessary for a complete FR monitoring diagnostic.

The recovery of 3D shape and camera motion for non-rigid scenes from single-camera video footage is a very important problem in computer vision. The low-rank shape model consists in regarding the deformations as linear combinations of basis shapes. Most algorithms for reconstructing the parameters of this model along with camera motion are based on three main steps. Given point tracks and the rank, or equivalently the number of basis shapes, they factorize a measurement matrix containing all point tracks, from which the camera motion and basis shapes are extracted and refined in a bundle adjustment manner. There are several issues that have not been addressed yet, among which, choosing the rank automatically and dealing with erroneous point tracks and missing data. We introduce theoretical and practical contributions that address these issues. We propose an implicit imaging model for non-rigid scenes from which we derive non-rigid matching tensors and closure constraints. We give a nonrigid Structure-From-Motion algorithm based on computing matching tensors over subsequences, from which the implicit cameras are extrated. Each non-rigid matching tensor is computed, along with the rank of the subsequence, using a robust estimator incorporating a model selection criterion that detects erroneous image points. Preliminary experimental results on real and simulated data show that our algorithm deals with challenging video sequences.