Antoine Dupret

High-current large-bandwidth photosensor on standard CMOS processes

On standard CMOS processes, basically two photosensors may be designed: photodiodes or vertical bipolar phototransistors. A trade-off must be found between the area of the sensor, its sensitivity and its bandwidth. In most designs, the high sensitivity of the sensor is a key point and led to choosing a phototransistor based solution. However this choice is made at the expense of the bandwidth of the sensor. For small currents, an analysis shows that it is mainly proportional to the base-emitter capacitance Cbe and to the collector current. Hence, in the case of a floating base bipolar and for a given current, the only way of reducing Cbe is to decrease the emitter area. On the other hand, the sensitivity is to be preserved. We have proposed and tested an original sensor based on the splitting of phototransistors. The basic idea is to use minimum size emitter bipolar transistors and to increase their collector-base junction perimeter. Thanks to this design, for a given sensor area, the bandwidth has been improved by a factor of 3 and the sensitivity has been preserved. This solution has been successfully used on an operational retina performing stochastic computations at video rates. In particular, thanks to our design, we have been able to successfully implement a 150 by 50 micrometer2 optoelectronic random generator providing up to 100,000 random variables per second.

Gaussian random number generation by differential detection of speckles

This PDF contains the communication Gaussian random number generation by differential detection of speckles.

An optoelectronic CMOS circuit implementing a simulated annealing algorithm

An original optoelectronic implementation of simulated annealing is presented. A compact and simple optical system provides a chip with arrays of independent random noise sources. The silicon chip is composed of a mesh of computing cells. Each cell includes both analog and digital circuits and includes two photosensors. A detailed analysis of this cell is given including a presentation of the design constraints. A 4/spl times/4-cells prototype chip was implemented in a 1 /spl mu/m CMOS digital technology and was successfully operated at 20000 iterations per second. The measurements and characterization of this chip made possible the successful design of a 600-cells chip also presented. These results demonstrate the video-rate application of simulated annealing to early vision tasks.

A 4000 Hz CMOS image sensor with in-pixel processing for light measurement and modulation

In this paper, we present a CMOS image sensor architecture coupling a spatial light modulator to a photodiode, for medical imaging based on acousto-optical coherence tomography with a digital holographic detection scheme. Our architecture is able to measure an interference pattern between a scattered beam transmitted through a scattering media and a reference beam, on an array with 16 μm pixel pitch, at 4000 Hz, which is compliant with correlation time of breast tissues. In-pixel processing allows generating from the incident light, a signal to polarize an embedded light modulator used to control the phase of the reflected beam. This reflected beam can then be focused on a region of interest of a scattering media, for therapy. The stacking of a photosensitive element with a spatial light modulator on the same device brings a significant robustness over the state of the art techniques such as perfect optical matching and reduced time delay in controlling light.

FPN Sources in Bolometric Infrared Detectors

Low manufacturing cost and ease of use favor spreading of 2-D bolometric infrared detector arrays over various application domains such as predictive maintenance, medical imaging, automotive industry, and security. The infrared detectors main figure of merit has long been the noise equivalent temperature difference (NETD), which sets the minimum temperature difference distinguishable from background noise at sensor output. However, while nowadays uncooled detectors have achieved sufficient NETD, fixed pattern noise (FPN) is indeed becoming a crucial figure of merit especially when the focal plane array (FPA) is not regulated by a thermoelectric cooler (TEC). In this paper, we study the various sources of dispersion of infrared bolometric detectors and their respective impact on FPN in the resulting image. We propose an analytical model to identify main sources of nonuniformity, and the confrontation of results with actual measurements leads to the ability of a highly accurate on-chip thermal drift compensation.

On adaptive pixel random selection for compressive sensing

Recently developed Compressive Sensing image sensor architectures tend to provide compact on-chip implementations to perform alternative acquisitions. On the other hand, the time of reconstruction generally limits possible applications taking advantage of those specific sensing schemes. This work proposes an entire Compressive Sensing system composed of an encoder (a dedicated imager top-level architecture) and a decoder (a reconstruction algorithm). The proposed system provides a compromise between the sensing scheme efficiency for relaxing on-chip constraints and the reconstruction complexity/quality. This system performs an adaptive block-based sensing, particularly well suited for video acquisition because of being combined with a fast inpainting based reconstruction algorithm. The simulation results show that compared to state of the art reconstructions and without important image degradation, the proposed reconstruction algorithm considerably reduces the computation time.

On compensating unknown pixel behaviors for image sensors with embedded processing

Some smart imagers embed image processing operations or Compressive Sensing at the focal plane level. This process introduces artifacts due to technology dispersion and unpredictable behaviors. This article presents a generic algorithm structure well suited for compensating block artifacts by appropriate post processing operations. The proposed restoration method is composed by a three steps loop: regularize the image, fit the model parameters and update the regularization coefficient (by reweighting the fidelity term). To further demonstrate the efficiency of the proposed generic technique, two specific case studies involving structured PRNU and multi-capture compressive sensing are presented with simulation results.

Model based on-chip 13bits ADC design dedicated to uncooled infrared focal plane arrays

This paper presents an on-chip 13 bits 10 M/S Analog to Digital Converter (ADC) specifically designed for infrared bolometric image sensor. Bolometric infrared sensors are MEMs based thermal sensors, which covers a large spectrum of infrared applications, ranging from night vision to predictive industrial maintenance and medical imaging. With the current move towards submicron technologies, the demand for more integrated, smarter sensors and microsystems has dramatically increased. This trend has strengthened the need of on-chip ADC as the interface between the analog core and the digital processing electronic. However designing an on-chip ADC dedicated to focal plane array raises many questions about its architecture and its performance requirements. To take into account those specific needs, a high level model has been developed prior to the actual design. In this paper, we present the trade-offs of ADC design linked to infrared key performance parameters and bolometric technology detection method. The original development scheme, based on system level modeling, is also discussed. Finally we present the actual design and the measured performances.

Co-integration of a smart CMOS image sensor and a spatial light modulator for real-time optical phase modulation

We present a CMOS light detector-actuator array, in which every pixel combines a spatial light modulator and a photodiode. It will be used in medical imaging based on acousto-optical coherence tomography with a digital holographic detection scheme. Our architecture is able to measure an interference pattern between a scattered beam transmitted through a scattering media and a reference beam. The array of 16 μm pixels pitch has a frame rate of several kfps, which makes this sensor compliant with the correlation time of light in biological tissues. In-pixel analog processing of the interference pattern allows controlling the polarization of a stacked light modulator and thus, to control the phase of the reflected beam. This reflected beam can then be focused on a region of interest, i.e. for therapy. The stacking of a photosensitive element with a spatial light modulator on the same chip brings a significant robustness over the state of the art such as perfect optical matching and reduced delay in controlling light.

Low complexity depth map extraction and all-in-focus rendering for close-to-the-pixel embedded platforms

Providing depth information along with images is a recurrent topic and many methods have been proposed in the literature, but most of them at the expense of heavy processing. With the development of new autonomous systems, smart sensors should be developed and algorithms simplified in order to embed processing closer to the signal acquisition. Among the reported depth extraction methods, depth from focus methods do not require heavy processing, they represent the most promising depth extraction process for close-to-the-pixel implementations. We developed low-processing-requirements depth-extraction methods based on bloc contrast analysis and compared them to a manually annotated reference ground truth and present obtained results along with results from similar works. Proposed methods are built on a sharpness criterion computed from horizontal and vertical gradients selected for their low complexity. Depth and confidence maps are then processed along horizontal and vertical directions and combined into unified depth and confidence maps. Two flavors of this process are presented: a high-resolution one for pixel-boundary precision and a low-resolution one that provides low-noise output and a drastic 4 to 1 memory requirement reduction. An error-removal method and smoothing method is also proposed to improve the final depth map. Finally, all-in-focus images are computed based on the previously processed depth maps. The algorithm output is a fully-focused RGB-D-C image.