An FPGA-based design for real-time Super Resolution Reconstruction

Since several decades, the camera spatial resolution is gradually increasing with the CMOS technology evolution. The image sensors provide more and more pixels, generating new constraints for the suitable optics. As an alternative, promising solutions propose Super Resolution (SR) image reconstruction to extend the image size without modifying the sensor architecture. Convincing state-of art studies demonstrate that these methods could even be implemented in real-time. Nevertheless, artifacts can be observed in highly textured areas of the image. In this paper, we propose a Local Adaptive Spatial Super Resolution (LASSR) method to fix this limitation. A real-time texture analysis is include…

Efficient smart-camera accelerator: A configurable motion estimator dedicated to video codec

Smart cameras are used in a large range of applications. Usually the smart cameras transmit the video or/and extracted information from the video scene, frequently on compressed format to fit with the application requirements. An efficient hardware accelerator that can be adapted and provide the required coding performances according to the events detected in the video, the available network bandwidth or user requirements, is therefore a key element for smart camera solutions. We propose in this paper to focus on a key part of the compression system: motion estimation. We have developed a flexible hardware implementation of the motion estimator based on FPGA component, fully compatible with…

An Efficient Hardware implementation of MQ Decoder of JPEG2000

International audience; JPEG2000 is an international standard for still images intended to overcome the shortcomings of the existing JPEG standard. Compared to JPEG image compression techniques, JPEG2000 standard has not only better not only has better compression ratios, but it also offers some exciting features. As it's hard to meet the real-time requirement of image compression systems by software, it is necessary to implement compression system by hardware. The MQ decoder of the JPEG2000 standard is an important bottleneck for real-time applications. In order to meet the real-time requirement we propose in this paper a novel architecture for a MQ decoder with high throughput which is co…

Analysis of compatibility between lighting devices and descriptive features using Parzen’s kernel: application to flaw inspection by artificial vision

We present a supervised method, developed for industrial inspections by artificial vision, to obtain an adapted combination of descriptive features and a lighting device. This method must be implemented under real-time constraints and therefore a minimal number of features must be selected. The method is based on the assessment of the discrimination power of many descriptive features. The objective is to select the combination of descriptive features and lighting system best able to discriminate flawed classes from defect-free classes. In the first step, probability densities are computed for flawed and defect-free classes and for each tested combination. The discrimination power of the fea…

Embedded System Study for Real Time Boosting Based Face Detection

This paper describes a study for a real time embedded face detection system. Recently, the boosting based face detection algorithms proposed by [(Viola, P and Jone, M, 2001); (Lienhart, R, et al., 2003)] have gained a lot of attention and are considered as the fastest accurate face detection algorithms today. However, the embedded implementation of such algorithms into hardware is still a challenge, since these algorithms are heavily based on memory access. A sequential implementation model is built showing its lack of regularity in time consuming and speed of detection. We propose a parallel implementation that exploits the parallelism and the pipelining in these algorithms. This implement…

Efficient smart-camera accelerator: an configurable motion estimator dedicated to video codec

International audience; Smart cameras are used in a large range of applications. Usually the smart cameras transmit the video or/and extracted information from the video scene, frequently on compressed format to fit with the application requirements. An efficient hardware accelerator that can be adapted and provide the required coding performances according to the events detected in the video, the available network bandwidth or user requirements, is therefore a key element for smart camera solutions. We propose in this paper to focus on a key part of the compression system: motion estimation. We have developed a flexible hardware implementation of the motion estimator based on FPGA componen…

Definition and Performance Evaluation of a Robust SVM Based Fall Detection Solution

We propose an automatic approach to detect falls in home environment. A Support Vector Machine based classifier is fed by a set of selected features extracted from human body silhouette tracking. The classifier is followed by filtering operations taking into account the temporal nature of a video. The features are based on height and width of human body bounding box, the user's trajectory with her/his orientation, Projection Histograms and moments of order 0, 1 and 2. We study several combinations of usual transformations of the features (Fourier Transform, Wavelet transform, first and second derivatives), and we show experimentally that it is possible to achieve high performance using a si…

Real-Time 3D Face Acquisition Using Reconfigurable Hybrid Architecture

Acquiring 3D data of human face is a general problem which can be applied in face recognition, virtual reality, and many other applications. It can be solved using stereovision. This technique consists in acquiring data in three dimensions from two cameras. The aim is to implement an algorithmic chain which makes it possible to obtain a three-dimensional space from two two-dimensional spaces: two images coming from the two cameras. Several implementations have already been considered. We propose a new simple real-time implementation based on a hybrid architecture (FPGA-DSP), allowing to consider an embedded and reconfigurable processing. Then we show our method which provides depth map of …

An efficient hardware implementation of MQ decoder of the JPEG2000

Abstract JPEG2000 is an international standard for still images intended to overcome the shortcomings of the existing JPEG standard. Compared to JPEG image compression techniques, JPEG2000 standard has not only better not only has better compression ratios, but it also offers some exciting features. As it’s hard to meet the real-time requirement of image compression systems by software, it is necessary to implement compression system by hardware. The MQ decoder of the JPEG2000 standard is an important bottleneck for real-time applications. In order to meet the real-time requirement we propose in this paper a novel architecture for a MQ decoder with high throughput which is comparable to tha…

Real-time flaw detection on a complex object: comparison of results using classification with a support vector machine, boosting, and hyperrectangle-based method

We present a classification work performed on industrial parts using artificial vision, a support vector machine (SVM), boost- ing, and a combination of classifiers. The object to be controlled is a coated heater used in television sets. Our project consists of detect- ing anomalies under manufacturer production, as well as in classi- fying the anomalies among 20 listed categories. Manufacturer speci- fications require a minimum of ten inspections per second without a decrease in the quality of the produced parts. This problem is ad- dressed by using a classification system relying on real-time ma- chine vision. To fulfill both real-time and quality constraints, three classification algorit…

Machine Learning VS Transfer Learning - Smart Camera Implementation for Face Authentication

The aim of this paper is to highlight differences between classical machine learning and transfer learning applied to low cost real-time face authentication. Furthermore, in an access control context, the size of biometric data should be minimized so it can be stored on a remote personal media. These constraints have led us to compare only lightest versions of these algorithms. Transfer learning applied on Mobilenet v1 raises to 85% of accuracy, for a 457Ko model, with 3680s and 1.43s for training and prediction tasks. In comparison, the fastest integrated method (Random Forest) shows accuracy up to 90% for a 7,9Ko model, with a fifth of a second to be trained and a hundred of microseconds …

Finding Invariants of Group Actions on Function Spaces, a General Methodology from Non-Abelian Harmonic Analysis

In this paper, we describe a general method using the abstract non-Abelian Fourier transform to construct “rich” invariants of group actions on functional spaces.

Detection and matching of curvilinear structures

We propose an approach to curvilinear and wiry object detection and matching based on a new curvilinear region detector (CRD) and a shape context-like descriptor (COH). Standard methods for local patch detection and description are not directly applicable to wiry objects and curvilinear structures, such as roads, railroads and rivers in satellite and aerial images, vessels and veins in medical images, cables, poles and fences in urban scenes, stems and tree branches in natural images, since they assume the object is compact, i.e. that most elliptical patches around features cover only the object. However, wiry objects often have no flat parts and most neighborhoods include both foreground a…

Hardware Implementation of a Configurable Motion Estimator for Adjusting the Video Coding Performances

International audience; Despite the diversity of video compression standard, the motion estimation still remains a key process which is used in most of them. Moreover, the required coding performances (bit-rate, PSNR, image spatial resolution, etc.) depend obviously of the application, the environment and the network communication. The motion estimation can therefore be adapted to fit with these performances. Meanwhile, the real time encoding is required in many applications. In order to reach this goal, we propose in this paper a hardware implementation of the motion estimator which enables the integer motion search algorithms to be modified and the fractional search and variable block siz…

An Efficient Hardware Implementation of Diamond Search Motion Estimation Based on CAL Dataflow Language

International audience

Definition of a Model-Based Detector of Curvilinear Regions

This paper describes a new approach for detection of curvilinear regions. These features detection can be useful for any matching based algorithm such as stereoscopic vision. Our detector is based on curvilinear structure model, defined observing the real world. Then, we propose a multi-scale search algorithm of curvilinear regions and we report some preliminary results.

High Efficiency Architecture of Half-Pel Motion Estimation for H.264 Video Coding

International audience

«Motion Estimation Accelerator with User Search Strategy in an RVC Context»

Motion estimation represents a key module in video compression. The RVC context requires proposing a flexible solution for motion estimation. According to the nature of the application, a full search is sometimes not suitable, hence, alternative fast/reduced solutions should be considered. This paper proposes a model and implementation of a flexible motion estimation engine, which can be configured to support any user-defined search strategy. Typically, the computational requirements of the search strategy can be traded with the RD-performance of the obtained video encoder. A CAL dataflow description of the accelerator is proposed so that it can be easily handled in the RVC context. An auto…

Fast prototyping of a SoC-based smart-camera: a real-time fall detection case study

International audience; Smart camera, i.e. cameras that are able to acquire and process images in real-time, is a typical example of the new embedded computer vision systems. A key example of application is automatic fall detection, which can be useful for helping elderly people in daily life. In this paper, we propose a methodology for development and fast-prototyping of a fall detection system based on such a smart camera, which allows to reduce the development time compared to standard approaches. Founded on a supervised classification approach, we propose a HW/SW implementation to detect falls in a home environment using a single camera and an optimized descriptor adapted to real-time t…

Texture Classification with Generalized Fourier Descriptors in Dimensionality Reduction Context: An Overview Exploration

In the context of texture classification, this article explores the capacity and the performance of some combinations of feature extraction, linear and nonlinear dimensionality reduction techniques and several kinds of classification methods. The performances are evaluated and compared in term of classification error. In order to test our texture classification protocol, the experiment carried out images from two different sources, the well known Brodatz database and our leaf texture images database.

SVM approximation for real-time image segmentation by using an improved hyperrectangles-based method

A real-time implementation of an approximation of the support vector machine (SVM) decision rule is proposed. This method is based on an improvement of a supervised classification method using hyperrectangles, which is useful for real-time image segmentation. The final decision combines the accuracy of the SVM learning algorithm and the speed of a hyperrectangles-based method. We review the principles of the classification methods and we evaluate the hardware implementation cost of each method. We present the combination algorithm, which consists of rejecting ambiguities in the learning set using SVM decision, before using the learning step of the hyperrectangles-based method. We present re…

Robust spatio-temporal descriptors for real-time SVM-based fall detection

Temozolomide and radiotherapy antitumor efficacy evaluation with magnetic resonance imaging and proton magnetic resonance spectroscopy in human glioma models in nude rats

International audience

Vineyard leaf roughness characterization by computer vision and cloud computing technics

International audience; In the context of vineyard leaf roughness analysis for precision spraying applications, this article deals with its characterization by computer vision and cloud computing techniques. The techniques merge feature extraction, linear or nonlinear dimensionality reduction techniques and several kinds of classification methods. Different combinations are processed and their performances compared in terms of classification error rate, in order to find the best association. However these combinations are hardly processed because of the lack of computing power and the prohibitive time consumption of the algorithms. To overcome these difficulties, we propose a solution: the …

Cost comparison of image rotation implantations on static and dynamic Reconfigurable FPGAs

FPGA components are widely used today to perform various algorithms (digital filtering) in real time. The emergence of Dynamically Reconfigurable (DR) FPGAs made it possible to reduce the number of necessary resources to carry out an image processing application (tasks chain). We present in this article an image processing application (image rotation) that exploits the FPGA 's dynamic reconfiguration feature. A comparison is undertaken between the dynamic and static reconfiguration by using two criteria, cost and performance criteria. For the sake of testing the validity of our approach in terms of Algorithm and Architecture Adequacy, we realized an AT40K40 based board ARDOISE.

Optimisation conjointe de la taille de stockage et des performances de modèles de classification pour l’authentification de visages

International audience

Real-time image segmentation for anomalies detection using SVM approximation

In this paper, we propose a method of implementation improvement of the decision rule of the support vector machine, applied to real-time image segmentation. We present very high speed decisions (approximately 10 ns per pixel) which can be useful for detection of anomalies on manufactured parts. We propose an original combination of classifiers allowing fast and robust classification applied to image segmentation. The SVM is used during a first step, pre-processing the training set and thus rejecting any ambiguities. The hyperrectangles-based learning algorithm is applied using the SVM classified training set. We show that the hyperrectangle method imitates the SVM method in terms of perfor…

An efficient hardware implementation of Diamond Search motion estimation using CAL dataflow language

Motion estimation represents a key module in video compression. The Reconfigurable Video Coding context (RVC) requires proposing a flexible solution for motion estimation. The motion estimation performance should be modified to fit with the user or the environment's constraints. Depending on the required performances fixed by the application, a full search is sometimes not suitable, hence, alternative fast/reduced solutions should be considered. In this paper, an efficient Diamond Search motion estimation, described in RVC-CAL actor language, is introduced. Starting from a high level description based CAL language, an automatic translation of the proposed CAL module to HDL is performed. Thi…

Classification Boundary Approximation by Using Combination of Training Steps for Real-Time Image Segmentation

We propose a method of real-time implementation of an approximation of the support vector machine decision rule. The method uses an improvement of a supervised classification method based on hyperrectangles, which is useful for real-time image segmentation. We increase the classification and speed performances using a combination of classification methods: a support vector machine is used during a pre-processing step. We recall the principles of the classification methods and we evaluate the hardware implementation cost of each method. We present our learning step combination algorithm and results obtained using Gaussian distributions and an example of image segmentation coming from a part …

Optimized Parallel Implementation of Face Detection based on GPU component

Display Omitted An algorithm for face detection has been implemented on CPU.An acceleration of this algorithm on GPU migration.Performance of GPU implementation shows the effectiveness of this implementation.Another optimization method on GPU are operated. Face detection is an important aspect for various domains such as: biometrics, video surveillance and human computer interaction. Generally a generic face processing system includes a face detection, or recognition step, as well as tracking and rendering phase. In this paper, we develop a real-time and robust face detection implementation based on GPU component. Face detection is performed by adapting the Viola and Jones algorithm. We hav…

Optimized spatio-temporal descriptors for real-time fall detection: comparison of support vector machine and Adaboost-based classification

We propose a supervised approach to detect falls in a home environment using an optimized descriptor adapted to real-time tasks. We introduce a realistic dataset of 222 videos, a new metric allowing evaluation of fall detection performance in a video stream, and an automatically optimized set of spatio-temporal descriptors which fed a supervised classifier. We build the initial spatio-temporal descriptor named STHF using several combinations of transformations of geometrical features (height and width of human body bounding box, the user’s trajectory with her/his orientation, projection histograms, and moments of orders 0, 1, and 2). We study the combinations of usual transformations of the…

Real-time flaw detection on complex part: Study of SVM and hyperrectangle based method

We present in this paper the study of two classifications methods used in order to control in real-time some industrials parts. We present the practical frame in which is made the operations, natures of the anomaly to be detected as well as the features extractions method. We tested two techniques of classification, with different algorithm complexities and performances. We compare the results obtained on various features spaces. We end by a combinatorial perspective of results of classification.

Real Time Robust Embedded Face Detection Using High Level Description

Face detection is a fundamental prerequisite step in the process of face recognition. It consists of automatically finding all the faces in an image despite the considerable variations of lighting, background, appearance of people, position/orientation of faces, and their sizes. This type of object detection has the distinction of having a very large intra-class, making it a particularly difficult problem to solve, especially when one wishes to achieve real time processing. A human being has a great ability to analyze images. He can extract the information about it and focus only on areas of interest (the phenomenon of attention). Thereafter he can detect faces in an extremely reliable way.…

A classification approach to prostate cancer localization in 3T Multi-Parametric MRI

International audience; Multiparametric-magnetic resonance imaging (mp-MRI) has demonstrated, in many studies, its potential in prostate cancer detection and analysis. We propose a supervised classification approach based on mp-MRI data base of 20 patients, in order to localize prostate cancer and to achieve a cartographic representation of the prostate voxels based on classification results. Proposed method provides a computer aided detection (CAD) software for prostatic cancer. For that, we have extracted varied features providing functional, anatomical and metabolic information helping the classifier to distinguish between three different classes ("Healthy", "Benign" and "Pathologic"). W…

<title>Face modeling: a real-time embedded implementation of a stereovision algorithm</title>

The problem to acquire 3D data of human face can be applied in face recognition, virtual reality, and many other applications. It can be solved using stereovision. This technique consistes in acquiring data in three dimensions from two cameras. The aim is to implement an algorithmic chain which makes it possible to obtain a three-dimensional space from two two-dimensional spaces: two images coming from the two cameras. Several implementations have already been considered. We propose a new simple realtime implementation, based on a multiprocessor approach (FPGA-DSP) allowing to consider an embedded processing. Then we show our method which provides depth map of face, dense and reliable, and …

Spatio-temporal descriptor for SVM and Adaboost based fall detection

International audience