Search

Capturing images has been increasingly popular in recent years, owing to the widespread availability of cameras. Images are essential in our daily lives because they contain a wealth of information, and it is often required to enhance images to obtain additional information. A variety of tools are available to improve image quality; nevertheless, they are also frequently used to falsify images, resulting in the spread of misinformation. This increases the severity and frequency of image forgeries, which is now a major source of concern. Numerous traditional techniques have been developed over time to detect image forgeries. In recent years, convolutional neural networks (CNNs) have received much attention, and CNN has also influenced the field of image forgery detection. However, most image forgery techniques based on CNN that exist in the literature are limited to detecting a specific type of forgery (either image splicing or copy-move). As a result, a technique capable of efficiently and accurately detecting the presence of unseen forgeries in an image is required. In this paper, we introduce a robust deep learning based system for identifying image forgeries in the context of double image compression. The difference between an image’s original and recompressed versions is used to train our model. The proposed model is lightweight, and its performance demonstrates that it is faster than state-of-the-art approaches. The experiment results are encouraging, with an overall validation accuracy of 92.23%.

Access to the floor plans of historical monuments over a time period is necessary in order to understand the architectural evolution and history. Such knowledge also helps to review (rebuild) the history by establishing connections between different events, persons and facts which were once part of the buildings. Since the two-dimensional plans do not capture the entire space, 3D modeling sheds new light on these unique archives and thus opens up great perspectives for understanding the ancient states of the monument. The first step towards generating the 3D model of the buildings and/or monuments is the wall detection inside the floor plan. Henceforth, the current work introduces a novel Versailles-FP dataset consisting Versailles Palace floor plan images and groundtruth in the form of wall masks regarding architectural developments during \(17^{th}\) and \(18^{th}\) century. The wall masks of the dataset are generated using an automated multi-directional steerable filters approach. The generated wall masks are then validated and corrected manually. We validate our approach of wall-mask generation in state-of-the-art modern datasets. Finally we propose a U-net based convolutional framework for wall detection. We have empirically shown that our U-net based method architecture achieves state-of-the-art results surpassing fully connected network based approach.

International audience; An effective model, which jointly captures shape and motion cues, for dynamic texture (DT) description is introduced by taking into account advantages of volumes of blurred-invariant features in three main following stages. First, a 3-dimensional Gaussian kernel is used to form smoothed sequences that allow to deal with well-known limitations of local encoding such as near uniform regions and sensitivity to noise. Second , a receptive volume of the Difference of Gaussians (DoG) is figured out to mitigate the negative impacts of environmental and illumination changes which are major challenges in DT understanding. Finally, a local encoding operator is addressed to construct a discriminative descriptor of enhancing patterns extracted from the filtered volumes. Evaluations on benchmark datasets (i.e., UCLA, DynTex, and DynTex++) for issue of DT classification have positively validated our crucial contributions.

International audience; This paper proposes a new local descriptor for action recognition in depth images using second-order directional Local Derivative Patterns (LDPs). LDP relies on local deriva- tive direction variations to capture local patterns contained in an image region. Our proposed local descriptor combines different directional LDPs computed from three depth maps obtained by representing depth sequences in three orthogonal views and is able to jointly encode the shape and motion cues. Moreover, we suggest the use of Sparse Coding-based Fisher Vector (SCFVC) for encoding local descriptors into a global representation of depth sequences. SCFVC has been proven effective for object recognition but has not gained much attention for action recognition. We perform action recognition using Extreme Learn- ing Machine (ELM). Experimental results on three public benchmark datasets show the effectiveness of the proposed approach.

International audience; In this paper, we propose a micro-macro feature combination approach for texture classification. The two disparate yet complementary categories of features are combined. By this way, Local Binary Pattern (LBP) plays the role of micro-structure feature extractor while the scattering transform captures macro-structure information. In fact, for extracting the macro-type features, coefficients are aggregated from three different layers of the scattering network. It is a handcrafted convolution network which is implemented by computing consecutively wavelet transforms and modulus non-linear operators. By contrast, in order to extract micro-structure features which are rotation-invariant, relatively robust to noise and illumination change, the completed LBP is utilized alongside the biologically-inspired filtering (BF) preprocessing technique. Overall, since the proposed framework can exploit the advantages of both feature types, its texture representation is not only invariant to rotation, scaling, illumination change but also highly discriminative. Intensive experiments conducted on many texture benchmarks such as CUReT, UIUC, KTH-TIPS-2b, and OUTEX show that our framework has a competitive classification accuracy.

International audience; In this paper, we present a novel descriptor for human action recognition, called Motion of Oriented Magnitudes Patterns (MOMP), which considers the relationships between the local gradient distributions of neighboring patches coming from successive frames in video. The proposed descriptor also characterizes the information changing across different orientations, is therefore very discriminative and robust. The major advantages of MOMP are its very fast computation time and simple implementation. Subsequently, our features are combined with an effective coding scheme VLAD (Vector of locally aggregated descriptors) in the feature representation step, and a SVM (Support Vector Machine) classifier in order to better represent and classify the actions. By experimenting on several common benchmarks, we obtain the state-of-the-art results on the KTH dataset as well as the performance comparable to the literature on the UCF Sport dataset.

International audience; Regarding texture features, Local-based methods such as Local Binary Pattern (LBP) and its variants are computationally efficient high-performing but sensitive to noise, and suffering global structure information loss. By contrast, filter-based counterparts, the Scattering Transform for instance, are tolerant to noise and translation but often lack of small local structure information. In this paper we propose an integration of those to take full advantages of both local and global features. In this way, LBP is used for extracting local features while the Scattering Transform feature plays the role of a global descriptor. In addition to the combination of these two state-of-the-art features, we further integrate a new preprocessing technique called biologically-inspired filtering (BF) as well as an efficient PCA classifier. Intensive experiments conducted on many texture benchmarks such as CUReT, UIUC, KTH-TIPS2b, and OUTEX show that our combined method not only outweighs each one which stands alone but also competes with state-of-the-art on the experimented datasets.

A new texture representation framework called statistical binary patterns (SBPs) is presented. It consists in applying rotation invariant local binary pattern operators ( LBP riu 2 ) to a series of moment images, defined by local statistics uniformly computed using a given spatial support. It can be seen as a generalisation of the commonly used complementation approach (CLBP), since it extends the local description not only to local contrast information, but also to higher order local variations. In short, SBPs aim at expanding LBP self-similarity operator from the local grey level to the regional distribution level. Thanks to a richer local description, the SBPs have better discrimination power than other LBP variants. Furthermore, thanks to the regularisation effect of the statistical moments, the SBP descriptors show better noise robustness than classical CLBPs. The interest of the approach is validated through a large experimental study performed on five texture databases: KTH-TIPS, KTH-TIPS 2b, CUReT, UIUC and DTD. The results show that, for the four first datasets, the SBPs are comparable or outperform the recent state-of-the-art methods, even using small support for the LBP operator, and using limited size spatial support for the computation of the local statistics. HighlightsWe extend the binary patterns from the pixel level to the local distribution level.We exploit moment images calculated from spatial support of the statistics.Statistical moments clearly improve the expressiveness and robustness of descriptor.

International audience; This work proposes an unsupervised jointalignment framework, referred to as ‘‘Gradient CorrelationCongealing,’’ which aligns an image ensemble bymaximizing a sum of gradient correlation coefficientfunction defined over all images. We, respectively, developtwo different formulations to optimize the objective functionregarding the role of ‘‘template.’’ While most existingface alignment methods suffer from outliers, e.g., occlusions,the proposed algorithms are able to align faces undergoingpartial occlusions. Moreover, our algorithms cancope with nonuniform illumination changes (even extremelydifficult ones), and also, they do not require anypredefined templates. We test the novel approaches againstfour typical joint alignment methods including Least-Squares Congealing, Learned-Miller Congealing, Lucas–Kanade entropy Congealing, and RASL using three challengingface databases: AR, Yale B, and LFW. Experimentalresults prove the efficiency of our approachesunder different conditions, especially when faces are partiallyoccluded, and the proposed algorithms perform muchbetter than all considered methods

International audience; This paper investigates the classification of photographic paper tex-tures using visual descriptors. Such classification is called fine grain due to the very low inter-class variability. We propose a novel image representation for photographic paper texture categorization, relying on the incorporation of a powerful local descriptor into an efficient higher-order model deviation where texture is represented by com-puting statistics on the occurrences of specific local visual patterns. We perform an evaluation on two different challenging datasets of photographic paper textures and show such advanced methods in-deed outperforms existing descriptors.

International audience; The intent of 3D-models classification is to find categories of similar objects according to their shapes. This task is a challenging and important problem in 3D-mining and shape processing. In this paper, we present a novel method to categorize 3D-objects based on view-based descriptors. The proposed method goes into two stages. The first stage corresponds to the training in which 3D-objects in the same category are processed and a set of representative 2D views is selected, The second stage corresponds to the labelling in which unknown objects are classified using a belief based classifier. The experimental results obtained on the Shrec07 datasets show that the system efficiently performs in categorizing 3D-models.

International audience; A new action model is proposed, by revisiting local binary patterns for dynamic texture models, applied on trajectory beams calculated on the video. The use of semi dense trajectory field allows to dramatically reduce the computation support to essential mo-tion information, while maintaining a large amount of data to ensure robustness of statistical bag of features action models. A new binary pattern, called Spatial Motion Pattern (SMP) is proposed, which captures self similarity of velocity around each tracked point(particle), along its trajectory. This operator highlights the geometric shape of rigid parts of moving objects in a video sequence. SMPs are combined with basic velocity in-formation to form the local action primitives. Then, a global representation of a space × time video block is provided by using hierarchical blockwise histograms, which allows to efficiently represent the action as a whole, while preserving a certain level of spa-tiotemporal relation between the action primitives. Inheriting from the efficiency and the invariance properties of both the semi dense tracker Video extruder and the LBP based representations, the method is designed for the fast computation of action descrip-tors in unconstrained videos. For improving both robustness and computation time in the case of high definition video, we also present an enhanced version of the semi dense tracker based on the so called super particles, which reduces the number of trajectories while improving their length, reliability and spatial distribution.

International audience; A new method for action recognition is proposed by revisit-ing LBP-based dynamic texture operators. It captures the similarity of motion around keypoints tracked by a realtime semi-dense point track-ing method. The use of self-similarity operator allows to highlight the geometric shape of rigid parts of foreground object in a video sequence. Inheriting from the efficient representation of LBP-based methods and the appearance invariance of patch matching method, the method is well designed for capturing action primitives in unconstrained videos. Action recognition experiments, made on several academic action datasets vali-date the interest of our approach.

International audience; Real-world face recognition systems require careful balancing of three concerns: computational cost, robustness, and discriminative power. In this paper we describe a new descriptor, POEM (Patterns of Oriented Edge Magnitudes) by applying a self-similarity based structure on oriented magnitudes and prove that it addresses all three criteria. Experimental results on the FERET databases show that POEM outperforms other descriptors when used with nearest neighbour classifiers. With LFW by combining POEM with GMMs and with multi-kernel SVMs, we achieve comparable results to the state of the art. Impressively, POEM is around 20 times faster than Gabor-based methods

International audience; Good face recognition system is one which quickly delivers high accurate results to the end user. For this purpose, face representation must be robust, discriminative and also of low computational cost in both terms of time and space. Inspired by recently proposed feature set so-called POEM (Patterns of Oriented Edge Magnitudes) which considers the relationships between edge distributions of different image patches and is argued balancing well the three concerns, this work proposes to further exploit patterns of both orientations and magnitudes for building more efficient algorithm. We first present novel features called Patterns of Dominant Orientations (PDO) which consider the relationships between "dominant" orientations of local image regions at different scales. We also propose to apply the whitened PCA technique upon both the POEM and PDO based representations to get more compact and discriminative face descriptors. We then show that the two methods have complementary strength and that by combining the two descriptors, one obtains stronger results than either of them considered separately. By experiments carried out on several common benchmarks, including both frontal and nonfrontal FERET as well as the AR datasets, we prove that our approach is more efficient than contemporary ones.

Poster session: Biometrics; International audience; Finding facial features respectively under expression and illumination variations is always a difficult problem. One popular solution for improving the performance of facial point localization is to use the spatial relation between facial feature positions. While existing algorithms mostly rely on the priori knowledge of facial structure and on a training phase, this paper presents an online approach without requirements of pre-defined constraints on feature distributions. Instead of training specific detectors for each facial feature, a generic method is first used to extract a set of interest points from test images. With a robust feature descriptor named Patterns Oriented Edge Magnitude (POEM), a smaller set of these points are picked as candidates. Then we apply a game-theoretic technique to select facial points from the candidates, while the global geometric properties of face are well preserved. The experimental results demonstrate that our method achieves satisfactory performance for face images under expression and lighting variations

Due to its wide variety of real-life applications, ranging from user-authentication (access control, ATM) to video surveillance and law enforcements, face recognition has been one of the most active research topics in computer vision and pattern recognition. Also, it has obvious advantages over other biometric techniques, since it is natural, socially well accepted, and notably non-intrusive. In reality, several reliable biometrics authentication techniques are available and widely used nowadays (such as iris or fingerprint), but they mostly rely on an active participation of the user. On the contrary, facial biometric demands very little cooperation from the user; thanks to this user-friendly capability, face recognition is said to be non-intrusive. Over the last decades, significant progress has been achieved in face recognition area. Since the seminal work of Turk and Pentland (Turk & Pentland, 1991), where the Principal Component Analysis (PCA) is proposed to apply to face images (Eigenfaces), more sophisticated techniques for face recognition appear, such as Fisherfaces (Belhumeur et al., 1997), based on linear discriminant analysis (LDA), Elastic Bunch Graph Matching (EBGM) (Wiskott et al., 1997), as well as approaches based upon Support Vector Machines (SVM) (Phillips, 1999), or Hidden Markov Models (HMM) (Nefian & III, 1998; Vu & Caplier, 2010b), etc. Nevertheless, face recognition, notably under uncontrolled scenarios, remains active and unsolved. Among many factors affecting the performance of face recognition systems, illumination is known to be one of the most significant. Indeed, it was proven, both theoretically (Moses et al., 1994) and experimentally (Adini et al., 1997) that image variation due to lighting changes is more significant than that due to different personal identities. In other words, the difference between two face images of the same individual taken under varying lighting conditions is larger than the difference between any two face images taken under the same lighting conditions, as illustrated in Fig. 1. Inspired by the great ability of human retina that enables the eyes to see objects in different lighting conditions, we present in this chapter a novel method of illumination normalization by simulating the performance of its two layers: the photoreceptors and the outer plexiform layer. Thus, we say the algorithm biologically inspired. The rest of the chapter is structured as follows: Section 2 briefly discusses the related work; Section 3 presents the model of retinal processing and its advantage. In Section 4, the 6

International audience; This paper addresses the question of computationally inexpensive yet discriminative and robust feature sets for real-world face recognition. The proposed descriptor named Patterns of Oriented Edge Magnitudes (POEM) has desirable properties: POEM (1) is an oriented, spatial multi-resolution descriptor capturing rich information about the original image; (2) is a multi-scale self-similarity based structure that results in robustness to exterior variations; and (3) is of low complexity and is therefore practical for real-time applications. Briefly speaking, for every pixel, the POEM feature is built by applying a self-similarity based structure on oriented magnitudes, calculated by accumulating a local histogram of gradient orientations over all pixels of image cells, centered on the considered pixel. The robustness and discriminative power of the POEM descriptor is evaluated for face recognition on both constrained (FERET) and unconstrained (LFW) datasets. Experimental results show that our algorithm achieves better performance than the state-of-the-art representations. More impressively, the computational cost of extracting the POEM descriptor is so low that it runs around 20 times faster than just the first step of the methods based upon Gabor filters. Moreover, its data storage requirements are 13 and 27 times smaller than those of the LGBP (Local Gabor Binary Patterns) and HGPP (Histogram of Gabor Phase Patterns) descriptors respectively.

International audience; In this paper we present a new architecture for face recognition with a single reference image, which separates the training process from the recognition process completely. In the training stage, by using a database containing various individuals, the spatial relations between face components are represented by two Hidden Markov Models, one of which models the similarity between face images coming from the same individual (within-subject similarities), and the other modelling inter-subject differences. This allows us to take a pair of face images, neither of which has been seen before, and determine whether or not they come from the same individual. Whilst other face-recognition HMMs use Maximum Likelihood, we have tested our approach using both Maximum Likelihood and MAP estimation, and find that MAP estimation provides better results. Importantly, the training database can be entirely separated from the gallery and test images: this means that adding new individuals to the system can be done without re-training. We present results based upon models trained on the FERET training dataset, and demonstrate that these give excellent recognition rates on both the FERET database itself and more impressively the unseen AR database. When compared to other HMM based face recognition techniques, our algorithm is of much lower complexity due to the small size of our observation sequence.

International audience; Performance of face recognition systems drop drastically when blur effect is present on facial images. In this paper, we propose a new approach for blurred face recognition. Our method is based on a measure of the level of blur introduced in the image using a no-reference blur metric. The face recognition process can be performed with any facial feature descriptor to allow the combination of alternative methods for overcoming data acquisition problems introduced in an image. To assess its efficiency, the approach has been applied with Gabor wavelets, Local Binary Patterns (LBP) and Local Phase Quantization (LPQ) facial descriptors on the FERET data-set. Experimental results clearly show the strength of this method at overcoming the problem caused by various forms of blur whatever the facial feature descriptor are implemented.

International audience; Illumination variations that occur on face images degrade the performances of face recognition systems. In this paper, we propose a novel method of illumination normalization based on retina modeling by combining two adaptive nonlinear functions and a Difference of Gaussians filter. The proposed algorithm is evaluated on the Yale B database, the Feret illumination database by using two face recognition methods: PCA and Local Binary Pattern (LBP). Experimental results show that the proposed method achieves very high recognition rates even for the most challenging illumination conditions. Our algorithm has also a low computational complexity