An area of particular importance in developing advanced imaging techniques concerns 3D motion measurement in small-scale mechatronics and automated microscopy. One major drawback is related to complex motion measurement with 6 degrees of freedom. In the proposed work, the extraction of unknown metrics such as focusing distance, in plane and out-of-plane positioning from digital holograms is performed including real‐time constraints. This work explores extended computer micro-vision capabilities offered by combining digital holographic microscopy (DHM) and last generation of deep learning algorithms such as Vision Transformer (ViT) networks. Our experiments show that the reconstruction in-focus distance can be predicted in DHM with a high accuracy using tiny modified architectures of deep ViT networks and convolutional neural networks (CNN). We compare ViT and Tiny ViT models with deep CNN usually used in digital holography such as VGG16, LeNet and AlexNet.
The real-time positioning of an object on a microscopic scale is a significant challenge and remains difficult to apply. Many traditional imaging techniques exist but their axial resolution and/or their measurement range is often limited. We develop a novel high‐profile technology based on three pillars to meet these challenges. Using digital holography, we determine the correct focus distance on a large scale. Secondly, a new generation transformer neural networks processes the hologram giving in real-time (~30 frames per seconds) a submicrometric axial resolution, exceeding therefore the diffraction limit of the depth of field. Finally, the spatial structuring of the object allows us a nanometric lateral positioning by classical techniques, which will be sped up by a machine learning technique. Such high frame rates enable real-time processing in many different application scenarios.
We develop a novel high‐profile application of machine learning techniques by elevating digital holography and sensing in robotics to a new level. The extraction of unknown metrics such as focusing distance and in plane positioning without full image restoration from digital holograms is performed by pre‐processing approach in space‐domain and/or in Fourier‐domain, including real‐time constraints. Measuring a single hologram, we successfully determine the axial distance of a complex object to the 10x microscope objective over a range of 100 µm with an accuracy of 1.25 µm. We apply a machine learning technique to the hologram to speed up tracking in the plane of the pseudo-periodic target position up to several tens of frames per second (fps). Such high frame rates enable real-time processing in many different application scenarios.
We propose a vision-based position sensor based on Digital Holography (DH) for in-plane and out-of-plane displacements measurement of a patterned plate with sub-pixel resolutions. DH is a lensless imaging principle using solid-state camera and/or spatial light modulators (SLM). Object scenes are generated or reconstructed numerically through wave propagation computations applied to a diffracted optical field recorded as an interferogram. The application of visual positioning to manipulation tasks in micro-robotics requires high accuracy and wide ranges of displacements that, unfortunately, are limited by finite depth-of-focus and fixed working distance of refractive imaging systems. Recently, we demonstrated that DH allows in-plane positioning of mobile targets ensuring nanometer resolutions at diverse working distances within a continuous range of more than 15 centimeters. By recording a set of digital holograms of a pseudo-periodic pattern fixed onto a moving target, images in phase and in intensity are restored by numerical reconstruction using Angular Spectrum Propagation methods by adjusting the reconstruction distance. A last step consists in performing a direct phase measurements of periodic pattern to reach nanometer resolutions. Three 2DFFT are required at minimum to extract the pattern position, which is time consuming if several hundred of holograms are recorded. We explore a new approach that consists to restore in-plane / out-of-plane position directly from the 2DFFT of the digital hologram without any need for image restitution. The proposed vision-based position sensor combines a 10 Mp CMOS camera and a SLM in order to perform a fine control of the interferometer reference arm.
This work presents micro-grids integrated to cell culture boxes. These grids allow the systematic registration of
the position of a zone observed by optical microscopy in a such way that it is possible to find it again easily for
new observations for instance after culture on drug injection. The position knowledge allows also the numerical
superimposition of recorded images in a common position reference system with a sub-pixel precision. It become
thus straightforward to perform a site by site analysis of the possible evolutions that may have occurred in the
biological medium.
A vision system is used for measuring in-plane target displacement, position and orientation. Pseudo-periodic
patterns fixed on the target forms a phase reference. Absolute position is determined with subpixel accuracy by
phase computations. Various position encoding designs are proposed for different displacement ranges and resolution.
Performances obtained are compared and discussed for both displacement and orientation measurements.
The capability to resolve position on depth ranges larger than the lens depth of focus is demonstrated.
This paper reports on a sub-pixel resolution vision approach for the characterization of in-plane rigid-body
vibration. It is based on digital processing of stroboscopic images of the moving part. The method involves a
sample preparation step, in order to pattern a periodic microstructure on the vibrating device, for instance by
focused ion beam milling. An image processing has then been developed to perform the optimum reconstruction
of this a priori known object feature. In-plane displacement and rotation are deduced simultaneously with a high
resolution (better than 0.01 pixel and 0.0005 rad. respectively). The measurement principle combines phase
measurements - that provide the high resolution - with correlation - that unwraps the phase with the proper
phase constants. The vibration modes of a tuning fork were fully characterized for the demonstration of the
method capabilities. Then the tuning fork was loaded with a tungsten wire sharpened in a sub-micrometer
tip for use in shear-force microscopy. The vibrations of the scanning probe were also characterized furnishing
representative data on its actual vibration amplitude. The technique could however be applied to many kinds of
micro-devices, for instance comb driven electrostatic actuators. For applications allowing the sample preparation,
the proposed methodology is more convenient than common interference methods or image processing techniques
for the characterization of the vibration modes, even for amplitudes in the nanometer range.
Optical Coherence Tomography is an emerging technique for biomedical diagnostic help. This is a non-invasive, high resolution, non-destructive mean for some optical biopsy. Since a few years new developments have been undergone in the field of OCT trying to functionalize OCT measurements. One of them is Spectroscopic OCT where simultaneous access to depth resolution as well as spectral features depth resolved in the media are obtained. These spectroscopic OCT system are mainly based on post processing of classical OCT signals what is time consuming and which add numerical noise. We propose an 'all optical' system for real-time direct display of depth-frequency analysis of media.
The sensor is dedicated to the detection of allergens. We use a biochemical reaction in the vicinity of the core of an optical fiber which modifies the propagation conditions of the optical wave by evanescent coupling. The detection involves a intrinsic optical fiber Fabry-Perot interferometer.
In interferometry sub-wavelength resolutions are achieved thanks to phase computations. The number of measurement points resolved per fringe depends on the signal to noise ratio and may vary from tens to billions. In this work, that principle of fringe interpolation by phase computation is applied to image processing for the construction of accurate position sensors based on vision. A micro-structured pattern is etched on a surface element and is used as a phase-reference signal similar to an interference fringe pattern. This surface element is fixed on the moving target of interest and is observed by a static vision system. Then target displacements are reconstructed and measured with a high accuracy by locating the phase-reference in the recorded images. In a first configuration in-plane position is retrieved with a resolution of about 10-2 pixel for horizontal and vertical coordinates and about 10-4 radian for the orientation. Thus nanometer displacements can be controlled by a diffractive optical system and micrometer-sized surface patterns. Furthermore, this method is self-calibrating since the phase-reference pattern dimensions are known accurately and then provide us with a size reference available in each recorded image. Different configurations allow the position measurement along the three-space directions while an interferometric set-up is able to locate the position versus the six degrees of freedom. Several measurements per second are performed with an up-to-date microcomputer thanks to the dedicated software developed.
KEYWORDS: Digital holography, Image sensors, Sensors, Modulation transfer functions, Holograms, Linear filtering, Optical filters, Image filtering, Digital recording, Spatial frequencies
The paper deals with the development of a micropositioning station that can be used in manufacturing systems, microfactories, AFM or SNOM microscopes. The central element of the station is a standing waves ultrasonic linear motor. It is a jump-stick-slip driving mechanism that can make longitudinal and lateral shiftings, and rotations in the horizontal plane. Its main merits are its ability to perform microscale displacements and to support heavy pre-loads. An adequate drive-amplifier is developed, allowing the control of the system. Even, a phase sensitive vision method is developed to sense the position and the heading angle with an accuracy better than 0.2 pixel i.e. 12 micrometers with an observed field of about 4 x 3 cm2. A personnel computer controls the whole system.
This paper addresses the localization problem in mobile robotics and microrobotics, which is necessary to be resolved for closed loop identification and control. An original method is developed which uses the image of a specific pattern and a signal processing method based on wavelet transforms. The pattern is made of two calibrated networks of straight lines perpendicular with respect to each other and which constitutes a reference phase excursion. With an adapted processing technique, a straight line equation is computed for each network characterizing its position as well as its orientation with a high accuracy. The intersection of those two perpendicular lines defines the mobile robot position and heading whose movement is studied. That technique leads to promising results: sub- micrometer accuracy in the position measurement.
White light interferometry can be seen as a multichannel process since each wavelength constitutes an independent information carrier. The simultaneous observation of different wavelengths allows the absolute evaluation of optical path differences (OPD). Two basic detection schemes are available which are chosen as a function of the application requirements. In the first one, all wavelengths are superimposed on the photodetector and we get a single composite output signal. Therefore, the OPD has to be scanned in order to detect the interference fringe pattern which appears only for OPD shorter than the light source correlation length. The second scheme consists to separate optically the different wavelengths for a parallel detection and to measure the absolute value of the OPD without any mechanical displacement of the interferometer. This paper explores the case of surface profilometry for the presentation of the latest proposed signal processing algorithms and to compare the capabilities and perspectives of those two approaches.
A common procedure of profilometry by means of white light interferometry is to scan one interferometer arm step by step. In this way, the intensity detected for each surface point reproduces the autocorrelogram of the light source, which is used for the determination of the absolute phase between a reference position and the zero optical path difference position. Phase changes due to reflection on the inspected surface produces a shift of the interference fringes with respect to the coherence envelope. If those phase changes vary from points to points, artifacts can be introduced in the profile reconstruction. We propose to measure simultaneously the interferometric phase and the shift of the interference fringes with respect to the coherence envelope. That processing is based on a wavelet transformation of the sampled light source correlograms and leads to complementary information that describes more completely the optical behavior of surfaces.
Phase shifting interferometry and white light interferometry are accurate techniques for surface analysis. These methods require an accurate scanning of the optical path difference for evaluating the phase at each surface point. Alternative techniques, avoiding the z surface scanning were recently proposed. They are based on a spectroscopic analysis of white light interferograms and were called Double Spectral Modulation (DSM). The interferometric set-up is associated to a diffraction grating and an image compression is performed. Thus absolute and unambiguous phase values are obtained along the spectral axis. This procedure allows nanometric resolution using two (or four) interferograms. In this paper we present a better approach to DSM which reduces by a factor of two the number of interferograms which must be analyzed. After data processing, a 3D image of the surface is reconstructed and any longitudinal or lateral scanning is avoided. This paper presents the principle and the data processing of the proposed method and experimental results as well.
Ambiguities can arise in measurements with interferometric profilers using monochromatic light in the presence of surface discontinuities larger than a quarter wavelength. In this case, surface profilometry can be performed by means of white light interferometry since the coherent function of the light source allows the identification of the fringe order. However, acquisition procedures and data treatments are more complicated than in the case of monochromatic phase-shifting interferometry. In this paper, the coherence function of the light source is considered as locally linear and new phase shifting algorithms can be derived, which are well adapted to white-light phase-shifting interferometry. First experimental results are presented.
The purpose of recent developments of profilometry by using white light interferometry is to provide new tools for the analysis of rough samples which when studied by monochromatic phase-shifting interferometry, may cause phase calculation ambiguities. The usual way to perform depth measurements by white light interferometry is to analyze the coherence-limited interference fringes while the optical path difference is scanned. The method proposed here does not use optical path difference scanning. A spectroscopic device is used instead to separate the interference intensities associated to each spectral component of the light source. Phase variations due to wavelength change are proportional to the optical path difference and allows depth measurement to be performed without axial scanning. The profile of one line of the inspected sample is obtained from only one 2D interferogram. In this 2D interferogram one direction corresponds to the inspected direction of the surface while the other one is the chromatic axis which allows phase to change with wavelength. Experimental results show the ability of the proposed method to obtain the profile of 1D surface with nanometric resolution.
A scanning optical profilometer without sample displacement along the optical axis has been developed. Light coming from a wide spectrum light source is focused onto the inspected surface through a Fresnel lens. Because of its dispersive properties, the foci are distributed along the optical axis as a function of wavelength. The reflected light is focused on a filtering pinhole like in confocal microscopy. The only wavelength exactly focused on the surface is transferred through this pinhole to the detection unit. The detected light is analyzed through a spectroscopic device and the surface height is deduced from the wavelength for which the light intensity is maximum. This punctual measure is repeated for each surface point by a lateral scanning of the surface. With respect to other focus sensing profilers, we replace a mechanical z displacement of the sample by a spectral inspection of the detected light. The setup and some experimental results are presented.
The aspect of manufactured objects gives rise to increasing attention in the industrial world. Because of the concurrence, aspect becomes an essential commercial criteria like technical performances or cost. To fulfill the necessary aspect control, industry tries to replace traditional visual control by systematic procedures and dedicated measurement apparatus. Our work deals with that domain of metrology. This paper describes a brightness measurement method working on rough surfaces with a skin profile (or waviness) from 50 to 500 micrometers deep. In such a case, the skin profile visibility is not due to height distribution of the surface and depth sensitivity of eye, but it is due to the amount of light that each point of the surface scatters toward the observer eye. Our eye discerns the skin profile through the contrast level of the image it catches. So the brightness (or the scattering properties) measurement of surfaces is essential for evaluating the aspect quality of product surfaces. Our method predicts the scattering properties of surfaces by the scattering theory developed by Beckmann and Spizzichino and from roughness measurement. Confocal microscopy is used for profiling surfaces within two different roughness ranges. The skin profile is measured with low magnification objectives and that data is used to compute the local slope of the surface. High magnification objectives allow us to measure microroughness superimposed on the local shape. After high pass filtering, the statistical roughness parameters are used to compute the reflection coefficient of the material from the Beckmann and Spizzichino model. Then the contrast of the image that an observer catches under any illumination and observation configurations is computed from the knowledge of both the skin profile and the brightness.
With a wide spectrum light source, the auto-correlation function of the emitted beam is nil except for very short values of the optical time delay. This property is used in an interferometric set-up to identify the zero optical path difference between the two interfering arms and to analyze surface profile. The proposed technique is based on a simple intensity detection and does not compute any phase calculation. Therefore, we are not concerned with the phase ambiguity problem associated to classical phase shifting interferometers. We first present the acquisition principle and discuss the operating conditions which guarantee a correct detection. Secondly, the experimental results show the nanometric resolution of the proposed technique. We finally explain the interest in combining this method with actual confocal or phase shifting profilometers.
The visual aspect of products meant for the general public is becoming an equally important criterium for their commercial success as their technical qualities and their cost. In most cases, the aesthetic quality of products is visually assessed by experienced people whose judgement is considered as a reference.This procedure has two major disadvantages: i) This visual estimation is entirely related to individual sensitivity and cannot be quantified. Only a comparative analysis is possible and requires a sufficient range of reference samples which are indispensable for this evaluation. ii) This visual evaluation concerns the global evaluation of the appearance and cannot dissociate with entire precision the roles of different parametres such as the color, the luminosity, the form and the shine of the surface. .. It is therefore rather difficult to point out possible modifications that could be brought to the process of fabrication in order to improve the aesthetic quality of the products. To be freed from the restrictions of visual inspection, the industrial world tries to endow itself with systematic procedures and instruments for physical measurement that evaluate the visual aspect of the surfaces. The study that we propose here belongs to this field of research, and applies more particularly to the measuring of the brightness of plastic pieces the surface of which presents an undulation of less than a millimetre deep (skin profile). In the first part of this study, we shallpropose a systematic analysis of visual aspect which four different and independant criteria defines. We shall explain why traditional methods for the measurement of scattering cannot be applied to such wavy surfaces. The material BRDF (Bidirectionnal Reflection Distribution Function) must be estimated from the analysis of their roughness by the means of established scattering models, such as that of Beckmann and Spizzichino. Secondly we shall introduce the procedure that we have developed in order to analyze the visual aspect of plastic samples, through confocal microscopy and image processing. A low magnification analysis allows one to measure the skin contours of the surfaces and to infer the local surface orientation which determines the exact values of the incidence and observation angles. A second analysis, with high magnification and an adequate image processing software, enables the extraction of the microroughness responsible for the scattering, and one can then evaluate the scattering properties of the material. Finally,the combination of the properties of scattering and of the local orientation of the surface allows the prediction of the degree of visibility of the skin profile of the surface. The statistical distribution of the measured parametres is also taken into account.
The modulation of the spectrum of a light beam is consider as a metrological tool.
In particular, double spectral modulation of a Super Luminiscent Laser Diode (SLD) is used to
analyze surface's profiles. Intensity and frequency modulation allows absolute measurements of
the surface without any auxiliary phase shifting. Depth and lateral resolution is determined by the
spectral resolution of the involved spectroscopic devices.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.