Nowadays, video surveillance systems are commonly found in most public and private spaces. These systems typically consist of a network of cameras that feed into a central node. However, the processing aspect is evolving toward distributed approaches, leveraging edge-computing. These distributed systems are capable of effectively addressing the detection of people or events at each individual node. Most of these systems, rely on the use of deep-learning and segmentation algorithms which enable them to achieve high performance, but usually with a significant computational cost, hindering real-time execution. This paper presents an approach for people detection and action recognition in the wild, optimized for running on the edge, and that is able to work in real-time, in an embedded platform. Human Action Recognition (HAR) is performed by using a Recurrent Neural Network (RNN), specifically a Long Short-Term Memory (LSTM). The input to the LSTM is an ad-hoc, lightweight feature vector obtained from the bounding box of each detected person in the video surveillance image. The resulting system is highly portable and easily scalable, providing a powerful tool for real-world video surveillance applications (in the wild and real-time action recognition). The proposal has been exhaustively evaluated and compared against other state-of-the-art (SOTA) proposals in five datasets, including four widely used (KTH, WEIZMAN, WVU, IXMAX) and a novel one (GBA) recorded in the wild, that includes several people performing different actions simultaneously. The obtained results validate the proposal, since it achieves SOTA accuracy within a much more complicated video surveillance real scenario, and using a lightweight embedded hardware.

In indoor localization there are applications in which the orientation of the agent to be located is as important as knowing the position. In this paper we present the results of the orientation estimation from a Local Positioning System based on Position Sensitive Device (PSD) sensors and the visible light emitted from the illumination of the room in which it is located. The orientation estimation will require that the PSD sensor receives signal from 2 light sources simultaneously. As will be shown in the article, the error determining the rotation angle of the agent with the on-board sensor is less than 0.2 degrees. The accuracy of the measurement has been evaluated experimentally in a 2.5 m high ceiling room over an area of 2.2 m2using geodetic measurement tools to establish the reference ground truth values.

The High connectivity among devices within the Internet-of-Things facilitates two-way flow of information throughout the infrastructure reaching homes and the consumers targeting broader energy goals. Our proposal encompasses consumers cooperating in response to utility supply conditions, i.e., electricity available from renewable sources. Such a smart and green community of consumers autonomously adapts its energy consumption by enabling a local aggregator to (1) integrate their demand into a common view and, (2) re-schedule the community demand given the renewable energy supply and the consumers’ demand time preferences. In this paper, we evaluate the developed scheduling algorithm using benchmark data to validate our proposal implementation over existent technology.

This paper shows a system and a method for indoor positioning based on the LED lighting installed on the environment. The proposal is based on the modulation of the visible light signal emitted in frequencies ranging from 50 to 200 kHz and detection by means of a PSD (Position Sensitive Detector) photodiode. The paper describes the mathematical modelisation of the proposed Indoor Positioning System (IPS), considering that the detector will move on a planar surface. Several synthetic results are shown for distances of 1 and 3 meters between the light source and detector. The positioning technique is based on the angle of arrival (AoA) captured from the LED lighting. Using genetic algorithm, we have obtained the model parameters for the detector (composed of a PSD and coupled lens). Different experimental tests have been carried out to determine the precision in positioning with the proposed system, considering a ceil lighting at 3 meters distance, with a coverage surface area of 3×3 m2. The average error results in the positioning of the detector, with low multipath (MP) effect, are below 1 cm in 80% of the tested points. Considering the presence of strong effects of multipath signals, the maximum positioning error remains below 4.5 cm.

This works presents advances in the characterization of multipath (MP) effects in Indoor Positioning Systems (IPS) using Position Sensitive Devices (PSD). The positioning errors introduced by MP are characterized while determining the Angle of Arrival (AoA) of an infrared signal (IR) to the PSD detector. From the obtained results it can be concluded that the MP effects in the positioning using AoA on PSDs have a reduced impact on the measurements.

Existing approaches for person re-identification are mainly based on creating distinctive representations or on learning optimal metrics. The achieved results are then provided in the form of a list of ranked matching persons. It often happens that the true match is not ranked first but it is in the first positions. This is mostly due to the visual ambiguities shared between the true match and other "similar" persons. At the current state, there is a lack of a study of such visual ambiguities which limit the re-identification performance within the first ranks. We believe that an analysis of the similar appearances of the first ranks can be helpful in detecting, hence removing, such visual ambiguities. We propose to achieve such a goal by introducing an unsupervised post-ranking framework. Once the initial ranking is available, content and context sets are extracted. Then, these are exploited to remove the visual ambiguities and to obtain the discriminant feature space which is finally exploited to compute the new ranking. An in-depth analysis of the performance achieved on three public benchmark data sets support our believes. For every data set, the proposed method remarkably improves the first ranks results and outperforms the state-of-the-art approaches.

Event-based strategies for control and estimation can offer great benefits to the field of indoor localisation. They achieve a more efficient use of resources, especially in network communications. In this paper we present the implementation of an event-based state estimator for indoor positioning of a mobile robot. The detection is carried out by a laser scanner attached to a mini-PC, that has not any knowledge of the robot's kinematics. This PC processes the laser data, obtains measurements of the robot position and then decides which samples are transmitted to a remote centre where the event-based estimation is carried out. The decision is made based on the distance between the actual measurement and the predicted position. We tested the performance of different event-based sampling methods and compared them with the periodic sampling. Additionally, we tested how the tuning the noise covariance matrices of the filter influences the estimation accuracy.

Person re-identification is an open and challenging problem in computer vision. Existing re-identification approaches focus on optimal methods for features matching (e.g., metric learning approaches) or study the inter-camera transformations of such features. These methods hardly ever pay attention to the problem of visual ambiguities shared between the first ranks. In this paper, we focus on such a problem and introduce an unsupervised ranking optimization approach based on discriminant context information analysis. The proposed approach refines a given initial ranking by removing the visual ambiguities common to first ranks. This is achieved by analyzing their content and context information. Extensive experiments on three publicly available benchmark datasets and different baseline methods have been conducted. Results demonstrate a remarkable improvement in the first positions of the ranking. Regardless of the selected dataset, state-of-the-art methods are strongly outperformed by our method.

Most of the open challenges in person re-identification arise from the large variations of human appearance and from the different camera views that may be involved, making pure feature matching an unreliable solution. To tackle these challenges state-of-the-art methods assume that a unique inter-camera transformation of features undergoes between two cameras. However, the combination of view points, scene illumination and photometric settings, etc., together with the appearance, pose and orientation of a person make the inter-camera transformation of features multi-modal. To address these challenges we introduce three main contributions. We propose a method to extract multiple frames of the same person with different orientation. We learn the pair wise feature dissimilarities space (PFDS) formed by the subspace of pair wise feature dissimilarities computed between images of persons with similar orientation and the subspace of pair wise feature dissimilarities computed between images of persons non-similar orientations. Finally, a classifier is trained to capture the multi-modal inter-camera transformation of pair wise images for each subspace. To validate the proposed approach we show the superior performance of our approach to state-of-the-art methods using two publicly available benchmark datasets.

People reidentification is one of the most challenging tasks in computer vision, and considerable efforts have been directed toward providing solutions to this problem. The existence of extensive camera networks and surveillance systems increases the amount of people images obtained, but, on the other hand, implies the need for new algorithms to enable reidentification of people captured by the cameras. There is no one optimal model that solves the entire problem, but a set of distinctive features can be used to help in the matching process. Our proposal consists of using the orientation of each person captured in the surveillance scene to considerably improve the reidentification process. An iterative algorithm maximizes the number of successful matches and speeds up the process. A comparison with other earlier relevant studies is presented using available datasets.

People counting systems are widely used in surveillance applications. In this paper, we present a solution to bidirectional people counting based on information provided by an overhead stereo system. Four fundamental aspects can be identified: the detection and tracking of human motion using an extended particle filter, the use of 3-D measurements in order to increase the system's robustness and a modified K-means algorithm to provide the number of hypotheses at each time, and, finally, trajectory generation to facilitate people counting in different directions. The proposed algorithm is designed to solve problems of occlusion, without counting objects such as shopping trolleys or bags. A processing ratio of around 30 frames/s is necessary in order to capture the real-time trajectory of people and obtain robust tracking results. We validated various test videos, achieving a hit rate between 95% and 99%, depending on the number of people crossing the counting area.

The emergence of Intelligent Transport Systems has led to the development of a specific communications standard carried out by the IEEE: Wireless Access in Vehicular Environments (WAVE). However, this standard does not address certain critical security and real time application constraints, such as the platooning guidance of transport units. This paper presents an original proposal of medium access control protocol based on Non Cooperative Cognitive Radio techniques to obtain a mechanism which complies with the requirements of real time communications. The mechanism also offers other advantages, such as avoiding signalling and the capacity to adapt to channel conditions. The solution is applied to the problem of units merging a convoy. Comparison results between the proposal and Slotted-Aloha are included.

A hardware solution is presented to obtain the eigenvalues and eigenvectors of a real and symmetrical matrix using field-programmable gate arrays (FPGAs). Currently, this system is used to compute the eigenvalues and eigenvectors in covariance matrices for applications in digital image processing that make use of the principal component analysis (PCA) technique. The proposed solution in this paper is based on the Jacobi method, but in comparison with other related works, it presents a different architecture that remarkably improves execution time, while reducing the number of consumed resources of the FPGA.

This work shows a modular architecture based on FPGA's to solve the eigenvalue problem according to the Jacobi method. This method is able to solve the eigenvalues and eigenvectors concurrently. The main contribution of this work is the low execution time compared with other sequential algorithms, and minimal internal FPGA consumed resources, mainly due to the fact of using the CORDIC algorithm. Two CORDIC modules have been designed to solve the trigonometric operations involved. A parallel CORDIC architecture is proposed as it is the best option to compute the eigenvalues with this method. Both CORDIC modules can work in rotation and vector mode. The whole system has been done in VHDL language, attempting to optimize the design.

This work shows different proposals to implement generic convolution with images from a vision camera. The developed algorithms present different methods to perform convolutions using VHDL language. The main problem of this operation is the high number of mathematical operations; numerous multiplications, additions and divisions are needed. This cause has led to create design electronic circuits, whose structure is based on internal FPGA resources (field programmable gate array). So the obtained result has been optimised in terms of speed and required resources in circuits. These kinds of generic operations are often used in artificial vision, as a pre-processing stage for images. This step precedes image processing.

This work deals with the implementation of the discrete 2D wavelet transform in VHDL. The proposed system has as advantages the segmentation of the wavelet algorithm in different processes and presents as novelty the control and addressing of input and output data. After introducing the key concepts of wavelet filtering the algorithm of wavelet transform is shown. Next, the functional simulation of the different blocks is described. The control algorithm capable to address the input data correctly and the state machine used to follow the evolution of the whole system are studied. One vertical processor and two horizontal processors in parallel are used to filter the image. Lastly, it becomes necessary a stage of output data reordering to obtain the image with the standard format. Results and applications are given.

This paper studies the multi-image calibration of a camera and how to carry it out in an automatic way making use of a projector. To prepare for multi-image calibration, it is necessary to calibrate the camera. Next, calibrate the projector using that camera. Then calibrate any camera with similar focal length to the previous one. The system doesn't need to incorporate any position platform and is completely automatic once it is disposed. It is based on the generation of images of a virtual pattern. This virtually generated pattern is projected on a fixed blackboard or a wall. Different on-line generated virtual patterns are visualized to carry out the auto-calibration. It is simple and can be repeated for a number of cameras that have an appropriate focal length. Also an improved method for focus of expansion using the same ideas is presented to obtain the principal point for a zoom camera.

Describes a vision-based commands generation system, by face movements, applied to the guidance of an electric wheelchair for handicapped people with severe disabilities. Using a 2D color face tracker and a fuzzy detector the system computes face movements of the user and, depending on them, some commands are generated to drive the wheelchair. The system is non-intrusive and it allows visibility and freedom of head movements. It is able to learn the face movements of the user in an automatic initial setup, working even for people of different races. It is adaptive and, therefore, robust to light and background changes in inside environments. We report on some experimental results of this kind of guidance and some conclusions about its performance.