In this paper, we study the performance invariance of convolutional neural networks when confronted with variable image sizes in the context of a more "wild steganalysis". First, we propose two algorithms and definitions for a fine experimental protocol with datasets owning "similar difficulty" and "similar security". The "smart crop 2" algorithm allows the introduction of the Nearly Nested Image Datasets (NNID) that ensure "a similar difficulty" between various datasets, and a dichotomous research algorithm allows a "similar security". Second, we show that invariance does not exist in state-of-the-art architectures. We also exhibit a difference in behavior depending on whether we test on images larger or smaller than the training images. Finally, based on the experiments, we propose to use the dilated convolution which leads to an improvement of a state-of-the-art architecture.

Autonomous vehicles are able to sense their environment and operate without human involvement. In particular, they have to be able to locate themselves. In order to achieve this task, geolocalisation files such as RINEX files are used. Due to the large amount of data, these files are very large and there is a big challenge in reducing their size for real-time navigation. However, this problem has not really been explored for RINEX files. In this paper, we propose an efficient method to losslessly compress RINEX files based on quadratic polynomial interpolation for autonomous vehicles navigation. Indeed, due to the orbits of the satellites around the earth, satellite signals can be modeled by parabolic curves. After interpolation of this model, prediction errors are computed as the differences between each original value of the satellite signal and its associated predicted value. Finally, these prediction errors are compressed using entropy coding. Experimental results show that our proposed method allows us to outperform the compression rate obtained by a previous state-of-the-art approach.

After 2015, CNN-based steganalysis approaches have started replacing the two-step machine-learning-based steganalysis approaches (feature extraction and classification), mainly due to the fact that they offer better performance.

For about 10 years, detecting the presence of a secret message hidden in an image was performed with an Ensemble Classifier trained with Rich features. In recent years, studies such as Xu et al. have indicated that well-designed convolutional Neural Networks (CNN) can achieve comparable performance to the two-step machine learning approaches. In this paper, we propose a CNN that outperforms the state-of-the-art in terms of error probability. The proposition is in the continuity of what has been recently proposed and it is a clever fusion of important bricks used in various papers. Among the essential parts of the CNN, one can cite the use of a pre-processing filterbank and a Truncation activation function, five convolutional layers with a Batch Normalization associated with a Scale Layer, as well as the use of a sufficiently sized fully connected section. An augmented database has also been used to improve the training of the CNN. Our CNN was experimentally evaluated against S-UNIWARD and WOW embedding algorithms and its performances were compared with those of three other methods: an Ensemble Classifier plus a Rich Model, and two other CNN steganalyzers.

In this paper, we present a new regularization paradigm for inverse based regularized image reconstruction techniques. These methods usually attempt to minimize a cost function expressed as the sum of a data-fitting term and a regularization term. The trade-off between both terms is determined by a weighting parameter that has to be set by the user since this trade-off is data dependent. In the approach we present here, we first concentrate on finding a set of eligible candidates for the data fitting term minimization and then select the most appropriate candidate according to the regularization criterion. The main advantage of this method is that it does not require any weighting parameter, and guarantees that no over-regularization can occur. We illustrate this method with a super-resolution reconstruction technique to show its efficiency compared to other competitive methods. Comparisons are carried out with simulated and real data.

The most effective superresolution methods proposed in the literature require precise knowledge of the so-called point spread function of the imager, while in practice its accurate estimation is nearly impossible. This paper presents a new superresolution method, whose main feature is its ability to account for the scant knowledge of the imager point spread function. This ability is based on representing this imprecise knowledge via a non-additive neighborhood function. The superresolution reconstruction algorithm transfers this imprecise knowledge to output by producing an imprecise (interval-valued) high-resolution image. We propose some experiments illustrating the robustness of the proposed method with respect to the imager point spread function. These experiments also highlight its high performance compared with very competitive earlier approaches. Finally, we show that the imprecision of the high-resolution interval-valued reconstructed image is a reconstruction error marker.

This paper presents a new iterative image super-resolution technique. Its main feature is its ability to account for a lack of knowledge of the so called Point Spread Function (PSF) of the imager. This ability is based on representing the ill-known PSF by mean of a maxitive kernel, i.e. an imprecise PSF. It also uses the nice properties of a fuzzy partition of the image plane for defining projection and back-projection operators whose particularity is to output interval-valued images instead of precise valued ones. Those operators transfer the imprecise knowledge on the PSF to the output imprecision. We propose some experiments illustrating the robustness of our approach with respect to the registration errors and its performance compared to very competitive earlier approaches.

This paper presents a new reconstruction operator to be used in a super-resolution scheme. Here, by reconstruction in super-resolution, we mean the back-projection operation, i.e. the way K low resolution (LR) images are aggregated to obtain a smooth high resolution (HR) image. Within this method, we replace the usual reconstruction procedure by a non-additive reconstruction operation based on the nice properties of fuzzy partitions. This non-additive reconstruction operator represents a convex family of usual additive reconstruction operators. The obtained reconstructed image is thus a convex family of usual reconstructed images. It allows the super-resolution method to be less sensitive to the choice of the reconstruction method. To make the reading of this method easier, it is presented with 1D signals. We present some experiments to illustrate the proved properties of this new operator.

The way catadioptric images are acquired implies that they present radial distortions. Therefore, classical processing may not be suitable. This statement will be illustrated by considering edge detection matter. Classical edge detectors usually consist in three steps : gradient computation, maximization and thresholding. The two lasts steps use pixels neighborhood concept. On the opposite of perspective images where pixel neighborhood is intuitive, catadioptric images present radial resolution changes. Then, the size and shape of pixel neighborhood have to be depending on pixel location. This article presents a new gradient estimation approach based on non-additive kernels. This technique is adapted to catadioptric images and also provides a natural threshold discarding the arbitrary thresholding step.

In this paper, we propose a new method to perform the first derivative estimation of a discrete intensity distribution. This approach is based on a non-additive aggregation process and provides an estimate of the gradient as intervals instead of single values. These intervals are used to threshold a gradient-based edge detection and therefore discard spurious detections due to noise.

This article describes an original method to track 2D templates in a video sequence. This approach is based on robust modelling of the interaction between pattern movement and grey level variations in the image using fuzzy bimodal rules. Movement estimation is performed by inverting these rules and polling the results in a quasicontinuous histogram. Some experimental results are given to illustrate the performance of this algorithm

This paper describes morphological tools that have been adapted to omnidirectional catadioptric images.

This paper deals with main apparent motion estimation using a fuzzy representation of a pixel's gray level. Rough histograms and possibility theory are used to provide an accurate estimation of the motion. The use of these theories allows us to partly cope with the limitations of classical motion estimation methods. This article explains the estimation process in detail and also presents some illustrative examples.

Applied statistics are widely used in pattern recognition and other computing applications to find the most likely value of a parameter. The use of classical empirical statistics is based upon assumption about normality of underlying density distribution of data. When the data is corrupted by contaminated noise, then classical tools are usually not robust enough and the estimation of the mode is biased. In this article, we propose to estimate the main mode of a distribution by means of a rough histogram and we show that this estimation is robust to contamination.