Gustavo Botelho de Souza

Deep Texture Features for Robust Face Spoofing Detection

Biometric systems are quite common in our everyday life. Despite the higher difficulty to circumvent them, nowadays criminals are developing techniques to accurately simulate physical, physiological, and behavioral traits of valid users, process known as spoofing attack. In this context, robust countermeasure methods must be developed and integrated with the traditional biometric applications in order to prevent such frauds. Despite face being a promising trait due to its convenience and acceptability, face recognition systems can be easily fooled with common printed photographs. Most of state-of-the-art antispoofing techniques for face recognition applications extract handcrafted texture features from images, mainly based on the efficient local binary patterns (LBP) descriptor, to characterize them. However, recent results indicate that high-level (deep) features are more robust for such complex tasks. In this brief, a novel approach for face spoofing detection that extracts deep texture features from images by integrating the LBP descriptor to a modified convolutional neural network is proposed. Experiments on the NUAA spoofing database indicate that such deep neural network (called LBPnet) and an extended version of it (n-LBPnet) outperform other state-of-the-art techniques, presenting great results in terms of attack detection.

Deep Boltzmann machines for robust fingerprint spoofing attack detection

Biometrie systems present some important advantages over the traditional knowledge-or possess-oriented identification systems, such as a guarantee of authenticity and convenience. However, due to their widespread usage in our society and despite the difficulty in attacking them, nowadays criminals are already developing techniques to simulate physical, physiological and behavioral traits of valid users, the so-called spoofing attacks. In this sense, new countermeasures must be developed and integrated with the traditional biometric systems to prevent such frauds. In this work, we present a novel robust and efficient approach to detect spoofing attacks in biometric systems (fingerprint-based ones) using a deep learning-based model: the Deep Boltzmann Machine (DBM). By extracting and working with high-level features from the original data, DBM can deal with complex patterns and work with features that can not be easily forged. The results show the proposed approach outperforms other state-of-the-art techniques, presenting high accuracy in terms of attack detection and allowing working with less labeled data.

Shape Analysis Using Multiscale Hough Transform Statistics

With the widespread proliferation of computers, many human activities entail the use of automatic image analysis. The basic features used for image analysis include color, texture, and shape. In this paper, we propose MHTS (Multiscale Hough Transform Statistics), a multiscale version of the shape description method called HTS (Hough Transform Statistics). Likewise HTS, MHTS uses statistics from the Hough Transform to characterize the shape of objects or regions in digital images. Experiments carried out on MPEG-7 CE-1 (Part B) shape database show that MHTS is better than the original HTS, and presents superior precision–recall results than some well-known shape description methods, such as: Tensor Scale, Multiscale Fractal Dimension, Fourier, and Contour Salience. Besides, when using the multiscale separability criterion, MHTS is also superior to Zernike Moments and Beam Angle Statistics (BAS) methods. The linear complexity of the HTS algorithm was preserved in this new multiscale version, making MHTS even more appropriate than BAS method for shape analysis in high-resolution image retrieval tasks when very large databases are used.

Roteiros abertos em filmes de busca

We will discuss in this text how characters of the films 33 (Kiko Goifman, 2002) and Olhe pra mim de novo (Kiko Goifman, Claudia Priscilla, 2009) handle the “risk of the real” (Comolli, 2008) in the course of these documentaries filmmaking, signaling to the construction of open screenplays which arrive at the time of shooting without prior definition of what will be recorded as image and sound in different planes, scenes, sequences. To do so, we follow methodological trails proposed by Aumont and Marie (2009) about “film analysis”, bringing them to the notions of “ search documentary” (Bernardet, 2005) and “documentary reading” (Roger Odin, 2012) to thus realize the search for family as a challenging issue in the films in question.

Deep Features Extraction for Robust Fingerprint Spoofing Attack Detection

Abstract Biometric systems have been widely considered as a synonym of security. However, in recent years, malicious people are violating them by presenting forged traits, such as gelatin fingers, to fool their capture sensors (spoofing attacks). To detect such frauds, methods based on traditional image descriptors have been developed, aiming liveness detection from the input data. However, due to their handcrafted approaches, most of them present low accuracy rates in challenging scenarios. In this work, we propose a novel method for fingerprint spoofing detection using the Deep Boltzmann Machines (DBM) for extraction of high-level features from the images. Such deep features are very discriminative, thus making complicated the task of forgery by attackers. Experiments show that the proposed method outperforms other state-of-the-art techniques, presenting high accuracy regarding attack detection.

Efficient Transfer Learning for Robust Face Spoofing Detection

Biometric systems are synonym of security. However, nowadays, criminals are violating them by presenting forged traits, such as facial photographs, to fool their capture sensors (spoofing attacks). In order to detect such frauds, handcrafted methods have been proposed. However, by working with raw data, most of them present low accuracy in challenging scenarios. To overcome problems like this, deep neural networks have been proposed and presented great results in many tasks. Despite being able to work with more robust and high-level features, an issue with such deep approaches is the lack of data for training, given their huge amount of parameters. Transfer Learning emerged as an alternative to deal with such problem. In this work, we propose an accurate and efficient approach for face spoofing detection based on Transfer Learning, i.e., using the very deep VGG-Face network, previously trained on large face recognition datasets, to extract robust features of facial images from the Replay-Attack spoofing database. An SVM is trained based on the feature vectors extracted by VGG-Face from the training images of Replay database in order to detect spoofing. This allowed us to work with such 16-layered network, obtaining great results, without overfitting and saving time and processing.

A Deep Boltzmann Machine-Based Approach for Robust Image Denoising

A Deep Boltzmann Machine (DBM) is composed of a stack of learners called Restricted Boltzmann Machines (RBMs), which correspond to a specific kind of stochastic energy-based networks. In this work, a DBM is applied to a robust image denoising by minimizing the contribution of some of its top nodes, called “noise nodes”, which often get excited when noise pixels are present in the given images. After training the DBM with noise and clean images, the detection and deactivation of the noise nodes allow reconstructing images with great quality, eliminating most of their noise. The results obtained from important public image datasets showed the validity of the proposed approach.

A 2D Deep Boltzmann Machine for Robust and Fast Vehicle Classification

The visual and automatic classification of vehicles plays an important role in the Transport Area. Besides of security issues, the monitoring of the type of traffic in streets and highways, as well the traffic dynamics over time, allows the optimization of use and of resources related to such public infrastructure. In this work we propose a novel method, called 2D-DBM, for robust and efficient automatic vehicle classification through color images based on a DBM (Deep Boltzmann Machine) combined with bilinear projections. While the DBM training allows a robust initialization of discriminative MLP (Multilayer Perceptron) neural network parameters, the bilinear projection technique can scale down the MLP dimensions, obtaining efficiency while preserving accuracy. The proposed method was assessed on the BIT-Vehicle database, a challenging dataset consisting of frontal images of vehicles collected in a real traffic environment, and compared with a CNN (Convolutional Neural Network) and a traditional DBM (without bilinear projection). The obtained results show that, while keeping the accuracy, the new method significantly reduced the network size and the processing time.

A Robust Restricted Boltzmann Machine for Binary Image Denoising

During the image acquisition process, some level of noise is usually added to the real data mainly due to physical limitations of the acquisition sensor, and also regarding imprecisions during the data transmission and manipulation. Therefore, the resultant image needs to be processed in order to attenuate its noise without loosing details. Machine learning approaches have been successfully used for image denoising. Among such approaches, Restricted Boltzmann Machine (RBM) is one of the most used technique for this purpose. Here, we propose to enhance the RBM performance on image denoising by adding a posterior supervision before its final denoising step. To this purpose, we propose a simple but effective approach that performs a fine-tuning in the RBM model. Experiments on public datasets corrupted by different levels of Gaussian noise support the effectiveness of the proposed approach with respect to some state-of-the-art image denoising approaches.

A Graph-Based Approach for Contextual Image Segmentation

Image segmentation is one of the most important tasks in Image Analysis since it allows locating the relevant regions of the images and discarding irrelevant information. Any mistake during this phase may cause serious problems to the subsequent methods of the image-based systems. The segmentation process is usually very complex since most of the images present some kind of noise. In this work, two techniques are combined to deal with such problem: one derived from the graph theory and other from the anisotropic filtering methods, both emphasizing the use of contextual information in order to classify each pixel in the image with higher precision. Given a noisy grayscale image, an anisotropic diffusion filter is applied in order to smooth the interior regions of the image, eliminating noise without loosing much information of boundary areas. After that, a graph is built based on the pixels of the obtained diffused image, linking adjacent nodes (pixels) and considering the capacity of the edges as a function of the filter properties. Then, after applying the Ford-Fulkerson algorithm, the minimum cut of the graph is found (following the min cut-max flow theorem), segmenting the object of interest. The results show that the proposed approach outperforms the traditional and well-referenced Otsus method.