Michel Paindavoine

Implementation of an RBF neural network on embedded systems: real-time face tracking and identity verification

This paper describes a real time vision system that allows us to localize faces in video sequences and verify their identity. These processes are image processing techniques based on the radial basis function (RBF) neural network approach. The robustness of this system has been evaluated quantitatively on eight video sequences. We have adapted our model for an application of face recognition using the Olivetti Research Laboratory (ORL), Cambridge, UK, database so as to compare the performance against other systems. We also describe three hardware implementations of our model on embedded systems based on the field programmable gate array (FPGA), zero instruction set computer (ZISC) chips, and digital signal processor (DSP) TMS320C62, respectively. We analyze the algorithm complexity and present results of hardware implementations in terms of the resources used and processing speed. The success rates of face tracking and identity verification are 92% (FPGA), 85% (ZISC), and 98.2% (DSP), respectively. For the three embedded systems, the processing speeds for images size of 288 /spl times/ 352 are 14 images/s, 25 images/s, and 4.8 images/s, respectively.

A 10 000 fps CMOS Sensor With Massively Parallel Image Processing

A high-speed analog VLSI image acquisition and pre-processing system has been designed and fabricated in a 0.35 ¿m standard CMOS process. The chip features a massively parallel architecture enabling the computation of programmable low-level image processing in each pixel. Extraction of spatial gradients and convolutions such as Sobel or Laplacian filters are implemented on the circuit. For this purpose, each 35 ¿m × 35 ¿m pixel includes a photodiode, an amplifier, two storage capacitors, and an analog arithmetic unit based on a four-quadrant multiplier architecture. The retina provides address-event coded output on three asynchronous buses: one output dedicated to the gradient and the other two to the pixel values. A 64 × 64 pixel proof-of-concept chip was fabricated. A dedicated embedded platform including FPGA and ADCs has also been designed to evaluate the vision chip. Measured results show that the proposed sensor successfully captures raw images up to 10 000 frames per second and runs low-level image processing at a frame rate of 2000 to 5000 frames per second.

Generalization of Canny---Deriche filter for detection of noisy exponential edge

This paper presents a generalization of the Canny-Deriche filter for ramp edge detection with optimization criteria used by Canny (signal-to-noise ratio, localization, and suppression of false responses). Using techniques similar to those developed by Deriche, we derive a filter which maximizes the product of the first two criteria under the constraint of the last one. The result is an infinite length impulse response filter which leads to a stable third-order recursive implementation. Its performance shows an increase of the signal-to-noise ratio in the case of blurred and noisy images, compared to the results obtained from Deriches filter.

High-speed smart camera with high resolution

High-speed video cameras are powerful tools for investigating for instance the biomechanics analysis or the movements of mechanical parts in manufacturing processes. In the past years, the use of CMOS sensors instead of CCDs has enabled the development of high-speed video cameras offering digital outputs, readout flexibility, and lower manufacturing costs. In this paper, we propose a high-speed smart camera based on a CMOS sensor with embedded processing. Two types of algorithms have been implemented. A compression algorithm, specific to high-speed imaging constraints, has been implemented. This implementation allows to reduce the large data flow (6.55 Gbps) and to propose a transfer on a serial output link (USB 2.0). The second type of algorithm is dedicated to feature extraction such as edge detection, markers extraction, or image analysis, wavelet analysis, and object tracking. These image processing algorithms have been implemented into an FPGA embedded inside the camera. These implementations are low-cost in terms of hardware resources. This FPGA technology allows us to process in real time 500 images per second with a 1280×1024 resolution. This camera system is a reconfigurable platform, other image processing algorithms can be implemented.

Small Sample Biometric Recognition Based on Palmprint and Face Fusion

Contactless biometrics provide high comfort and hygiene in person recognition. Because of this, such systems are better accepted by the general public. This paper proposes an adaptive, contactless, biometric system which combines two modalities: palmprint and face. The processing chain has been designed to overcome embedded system constraints and small sample set problem: after a palmprint is extracted from a hand image, Gabor filters are applied to both the palmprint and face in order to extract parameters, which are then used for classification. Fusion possibilities are also discussed and tested using a multimodal database of 130 people designed by the authors. High recognition performance has been obtained by respecting embedded system context, with palmprint only and with fusion of palmprint and face: recognition rates of respectively 96.39% and 98.85% are achieved using only 2 samples per modality. Therefore this preliminary study shows the feasibility of a robust and efficient multimodal hardware biometric system.

Simple Fingerprint Minutiae Extraction Algorithm Using Crossing Number On Valley Structure

In fingerprint recognition system, performance of fingerprint feature extraction algorithm is important. We use visual analysis to evaluate this performance. 100 respondents fill a questionnaire consisting of 30 images from fingerprint feature extraction process. We get 12,3 % minutiae points missed by this algorithm. With BOZORTH3 minutiae matching algorithm, the distribution of matching score of 80-fingerprint images are presented and we obtain EER 5.89 % at threshold value 180.

Computational Evidence That Frequency Trajectory Theory Does Not Oppose But Emerges From Age‐of‐Acquisition Theory

According to the age-of-acquisition hypothesis, words acquired early in life are processed faster and more accurately than words acquired later. Connectionist models have begun to explore the influence of the age/order of acquisition of items (and also their frequency of encounter). This study attempts to reconcile two different methodological and theoretical approaches (proposed by Lambon Ralph & Ehsan, 2006 and Zevin & Seidenberg, 2002) to age-limited learning effects. The current simulations extend the findings reported by Zevin and Seidenberg (2002) that have shown that frequency trajectories (FTs) have limited and specific effects on word-reading tasks. Using the methodological framework proposed by Lambon Ralph and Ehsan (2006), which makes it possible to compare word-reading and picture-naming tasks in connectionist networks, we were able to show that FT has a considerable influence on age-limited learning effects in a picture naming task. The findings show that when the input-output mappings are arbitrary (simulating picture naming tasks), the links formed by the network become entrenched as a result of early experience and that subsequent variations in frequency of exposure of the items have only a minor impact. In contrast, when the mappings between input-output are quasi-systematic or systematic (simulating word-reading tasks), the training of new items was generalized and resulted in the suppression of age-limited learning effects. At a theoretical level, we suggest that FT, which simultaneously takes account of time and the level of exposure across time, represents a more precise and modulated measure compared with the order of introduction of the items and may lead to innovative hypotheses in the field of age-limited learning effects.

Development of a fast panoramic face mosaicking and recognition system

We present some development results on a system that per- forms mosaicking of panoramic faces. Our objective is to study the fea- sibility of panoramic face construction in real time. To do so, we built a simple acquisition system composed of five standard cameras, which together can take simultaneously five views of a face at different angles. Then, we chose an easily hardware-achievable algorithm, consisting of successive linear transformations, in order to compose a panoramic face from these five views. The method has been tested on a large number of faces. In order to validate our system, we also conducted a preliminary study on panoramic face recognition, based on the principal-component method. Experimental results show the feasibility and viability of our sys- tem. This allows us to envisage later a hardware implementation. We also are considering using our system for other applications, such as human expression categorization and fast 3-D face reconstruction.

Automatic hardware implementation tool for a discrete Adaboost-based decision algorithm

We propose a method and a tool for automatic generation of hardware implementation of a decision rule based on the Adaboost algorithm. We review the principles of the classification method and we evaluate its hardware implementation cost in terms of FPGAs slice, using different weak classifiers based on the general concept of hyperrectangle. The main novelty of our approach is that the tool allows the user to find automatically an appropriate tradeoff between classification performances and hardware implementation cost, and that the generated architecture is optimized for each training process. We present results obtained using Gaussian distributions and examples from UCI databases. Finally, we present an example of industrial application of real-time textured image segmentation.

Measurement of the Motion of Fertilizer Particles Leaving a Centrifugal Spreader Using a Fast Imaging System

Although mechanically simple, centrifugal spreaders used for mineral fertilization involve complex physics that cannot be fully characterized at the present time. We are developing sensors to evaluate the spatial distribution of the fertilizer on the ground based on the measurement of initial flight conditions of fertilizer granules after their ejection by the spreading disk. The techniques developed are based on the analysis of images of the area around the disk showing the granule ejection. A high resolution – low cost imaging system for the analysis of high speed particle projection developed for this specific purpose is presented in this paper. The system, based on a camera and a sequence of flashes, is used to characterize the centrifugal spreading of fertilizer particles ejected at speeds of approximately 30m s−1. It automatically computes the direction of ejection and velocity of each granule observed in the image. Multi-exposure images collected with the camera installed perpendicular to the output flow of granules are analyzed to estimate the trajectories of the fertilizer granules, using different motion estimation methods.