In this work, we propose an image authentication method based on image matting. We demonstrate that an infinitesimal confidence interval computed from an $\alpha$-matte of a reference image is able to expose manipulated images at the pixel level. The corresponding critical bounds are infinitesimally conditional on the spatial and intensity image affinity which is determined by a constrained soft matting. A single reconstruction parameter is embedded in the image rather than embedding a watermark which is known to cause image distortions. The test images showed a high visual quality achieving a distinctive peak signal-to-noise ratio. The results indicated also a higher sensitivity of the reconstructed image residuals regarding the bounds of the infinitesimal confidence interval due to the broken affinity. Moreover, the approach implements a visual assessment for image authentication.

With the increasing development of advanced unmanned aerial vehicles (UAVs), communication between operators and these intelligent systems is becoming more stressful. For the safety of UAV flights, automatic psychological stress detection is becoming a key research topic for successful missions. Stress can be reliably estimated via some biological markers which are not appropriate in many cases of human-machine-interaction setups. In this article, we propose a non-intrusive deep learning-based stress level estimation approach. The goal is to identify the region where the operator's emotional state projects in the space defined by the latent dimensional emotions of arousal and valence since the stress region is well delimited in this space. The proposed multimodal approach uses sequential temporal CNN and LSTM with an Attention Weighted Average layer in the vision modality. As a second modality, we investigate local and global descriptors such as Mel-frequency cepstral coefficients, i-vector embeddings as well as Fisher-vector encodings. The multimodal-fusion approach uses a strategy referred to as “late-fusion” that involves the combination of unimodal model outputs as inputs of the decision engine. Since we have to deal with more naturalistic behavior in operator-machine interaction contexts, the One minute Gradual Emotion Challenge dataset was used for predictive model validation.

This work presents a Bayesian statistical approach to the multimodal change detection (CD) problem in remote sensing imagery. More precisely, we formulate the multimodal CD problem in the unsupervised Markovian framework. The main novelty of the proposed Markovian model lies in the use of an observation field built up from a pixel pairwise modeling and on the bitemporal heterogeneous satellite image pair. Such modeling allows us to rely instead on a robust visual cue, with the appealing property of being quasi-invariant to the imaging (multi-) modality. To use this observation cue as part of a stochastic likelihood model, we first rely on a preliminary iterative estimation technique that takes into account the variety of the laws in the distribution mixture and estimates the parameters of the Markovian mixture model. Once this estimation step is completed, the Maximum a posteriori (MAP) solution of the change detection map, based on the previously estimated parameters, is then computed with a stochastic optimization process. Experimental results and comparisons involving a mixture of different types of imaging modalities confirm the robustness of the proposed approach.

Because of speckle noise, estimating extended target scattering properties from Single Look Complex PolSAR data is a difficult problem. Speckle filtering is aiming at reducing noise within extended targets while preserving point target, polarimetric signatures and meaningful details. However, optimal speckle filtering requires some knowledge about the underlying target properties. Also, using all the terms of the quadpol polarimetric matrix in order to drive speckle filtering is still an open problem. Deep Learning techniques have been successful in tackling difficult computer vision tasks and show promising results when applied to SAR and PolSAR data. The goal of this paper is to investigate the use of Convolutional Neural Networks (CNN) in order to extract information from the full quadpol covariance matrix and help improve speckle filtering. Experiments are conducted using simulated PolSAR data and results show that we can potentially recover information from the off-diagonal terms.

In this paper, we propose a new multimodal change detection in remote sensing. The proposed method is based on a projection of the two multisensor satellite images to a common feature space, in which the two heterogeneous images share the same statistical properties and on which any classical monomodal change detection methods can be applied. This transformation of the before and after images is mainly based on a Multidimensional Scaling(MDS) representation which can be also viewed as a de-texturing approach of the two multisource images. Experimental results involving different types of imaging techniques confirm the reliability of the proposed approach.

Speckle removal from single-channel and multi-dimensional SAR remains a difficult problem. In this paper, we are investigating the use of a Convolutional Neural Network (CNN), previously applied to the Super-Resolution problem, for speckle removal. Because speckle noise statistics is signal dependent, we are training the neural network on the residual image formed by the ratio of the observed intensity to the local mean. Preliminary results on SAR and PolSAR show the potential of CNN to retrieve lost image details.

Extracting and identifying objects in very high resolution imagery has been a popular research topic in remote sensing. Since the beginning of this decade, deep learning techniques have revolutionized computer vision providing significant performance gains compared to traditional “shallow” techniques in various challenging vision problems. The training of deep neural networks usually requires very large training datasets. The advantage of using deep features is to exploit already trained Convolutional Neural Networks (CNN) in order to produce high level features without the burden of having to train a CNN from scratch. In this paper, we are investigating the use of deep features for the detection of small objects (cars and individual trees) in high resolution Pleiades imagery. Preliminary results show good detection performance and are very encouraging for future applications.

Water surface extraction using satellite images proves to be of great importance due to its utility in several applications such as land use, floods management and monitoring. Among the wide range of sensors orbiting around the earth, Synthetic Aperture Radar (SAR) proves to be a very effective tool in this context due to its robustness to unfavorable weather conditions and its cloud penetrating capabilities. This paper presents a novel rivers extraction method from SAR images mainly based on the combination of a local texture measurement and global knowledge associated to the shape of the object of interest. A local texture measurement is first computed for every pixel of the image to extract homogeneous surfaces, then a mathematical morphology operator is applied to attenuate noise generated by speckle characterizing SAR images. Finally, the surface occupied by the object of interest is compared to the surface associated to the smallest rectangle that encloses this object in order to separate rivers from lakes in the image. The proposed approach was tested on SAR images acquired by RADARSAT-2 satellite from numerous regions of Canada. Our experimental results demonstrate that the proposed approach is robust and effective.

Automatic facial expression analysis systems are aiming towards the application of computer vision techniques in human computer interaction, emotion analysis, and even medical care via a space mapping between the continuous emotion and a set of discrete expression categories. The main difficulty with these systems is the inherent problem of facial alignment due to person-specific appearance. Beside the facial representation problem, the same displayed facial expression may vary differently across humans; this can be true even for the same person in different contexts. To cope with these variable factors, we introduce the concept of prototype-based model as anchor modeling through a SIFT-flow registration. A set of prototype facial expression models is generated as a reference space of emotions on which face images are projected to generate a set of registered faces. To characterize the facial expression appearance, oriented gradients are processed on each registered image. We obtained the best results 87% with the person–independent evaluation strategy on JAFFE dataset (7-class expression recognition problem), and 83% on the complex setting of the GEMEP-FERA database (5-class problem).

Automatic facial expression analysis is the most commonly studied aspect of behavior understanding and human-computer interface. The main difficulty with facial emotion recognition system is to implement general expression models. The same facial expression may vary differently across humans; this can be true even for the same person when the expression is displayed in different contexts. These factors present a significant challenge for the recognition task. The method we applied, which is reminiscent of the “baseline method”, utilizes dynamic dense appearance descriptors and statistical machine learning techniques. Histograms of oriented gradients (HoG) are used to extract the appearance features by accumulating the gradient magnitudes for a set of orientations in 1-D histograms defined over a size-adaptive dense grid, and Support Vector Machines with Radial Basis Function kernels are the base learners of emotions. The overall classification performance of the emotion detection reached 70% which is better than the 56% accuracy achieved by the “baseline method” presented by the challenge organizers.