Mitigating non-Lambertian surfaces issues in Stereo Matching with Neural Radiance Fields

Depth estimation from images has long been regarded as a preferable alternative compared to expensive and intrusive active sensors, such as LiDAR and ToF. The topic has attracted the attention of an increasingly wide audience thanks to the great amount of application domains, such as autonomous driving, robotic navigation and 3D reconstruction. Among the various techniques employed for depth estimation, stereo matching is one of the most widespread, owing to its robustness, speed and simplicity in setup. Recent developments has been aided by the abundance of annotated stereo images, which granted to deep learning the opportunity to thrive in a research area where deep networks can reach state-of-the-art sub-pixel precision in most cases. Despite the recent findings, stereo matching still begets many open challenges, two among them being finding pixel correspondences in presence of objects that exhibits a non-Lambertian behaviour and processing high-resolution images. Recently, a novel dataset named Booster, which contains high-resolution stereo pairs featuring a large collection of labeled non-Lambertian objects, has been released. The work shown that training state-of-the-art deep neural network on such data improves the generalization capabilities of these networks also in presence of non-Lambertian surfaces. Regardless being a further step to tackle the aforementioned challenge, Booster includes a rather small number of annotated images, and thus cannot satisfy the intensive training requirements of deep learning. This thesis work aims to investigate novel view synthesis techniques to augment the Booster dataset, with ultimate goal of improving stereo matching reliability in presence of high-resolution images that displays non-Lambertian surfaces.

Surface reconstruction using multiple light sources and perspective projection

This paper presents a method to recover 3D geometry of Lambertian surfaces by using multiple images taken from the same view point and with the scene illuminated from different positions. This approach differs from Stereo Photometry in that it considers the light source at a finite distance from the object and the perspective projection in image formation. The proposed model allows local solution and recovery of 3D coordinates, in addition to surface orientation. A procedure to calibrate the light sources is also presented. Results of the application of the algorithm to synthetic images are shown.

Metodologie di Corrispondenza Stereo Basate su Deep Learning per Superfici Altamente Riflettenti e Trasparenti: Dataset e Architettura

Nell’ambito della Stereo Vision, settore della Computer Vision, partendo da coppie di immagini RGB, si cerca di ricostruire la profondità della scena. La maggior parte degli algoritmi utilizzati per questo compito ipotizzano che tutte le superfici presenti nella scena siano lambertiane. Quando sono presenti superfici non lambertiane (riflettenti o trasparenti), gli algoritmi stereo esistenti sbagliano la predizione della profondità. Per risolvere questo problema, durante l’esperienza di tirocinio, si è realizzato un dataset contenente oggetti trasparenti e riflettenti che sono la base per l’allenamento della rete. Agli oggetti presenti nelle scene sono associate annotazioni 3D usate per allenare la rete. Invece, nel seguente lavoro di tesi, utilizzando l’algoritmo RAFT-Stereo [1], rete allo stato dell’arte per la stereo vision, si analizza come la rete modifica le sue prestazioni (predizione della disparità) se al suo interno viene inserito un modulo per la segmentazione semantica degli oggetti. Si introduce questo layer aggiuntivo perché, trovare la corrispondenza tra due punti appartenenti a superfici lambertiane, risulta essere molto complesso per una normale rete. Si vuole utilizzare l’informazione semantica per riconoscere questi tipi di superfici e così migliorarne la disparità. È stata scelta questa architettura neurale in quanto, durante l’esperienza di tirocinio riguardante la creazione del dataset Booster [2], è risultata la migliore su questo dataset. L’obiettivo ultimo di questo lavoro è vedere se il riconoscimento di superfici non lambertiane, da parte del modulo semantico, influenza la predizione della disparità migliorandola. Nell’ambito della stereo vision, gli elementi riflettenti e trasparenti risultano estremamente complessi da analizzare, ma restano tuttora oggetto di studio dati gli svariati settori di applicazione come la guida autonoma e la robotica.

Digital surface model, hillshade and Lambertian reflectance model of the rock glaciers Oelgrube and Aeusseres Hochebenkar (Oetztal Alps, Tyrol, Austria)

With full-waveform (FWF) lidar systems becoming increasingly available from different commercial manufacturers, the possibility for extracting physical parameters of the scanned surfaces in an area-wide sense, as addendum to their geometric representation, has risen as well. The mentioned FWF systems digitize the temporal profiles of the transmitted laser pulse and of its backscattered echoes, allowing for a reliable determination of the target distance to the instrument and of physical target quantities by means of radiometric calibration, one of such quantities being the diffuse Lambertian reflectance. The delineation of glaciers is a time-consuming task, commonly performed manually by experts and involving field trips as well as image interpretation of orthophotos, digital terrain models and shaded reliefs. In this study, the diffuse Lambertian reflectance was compared to the glacier outlines mapped by experts. We start the presentation with the workflow for analysis of FWF data, their direct georeferencing and the calculation of the diffuse Lambertian reflectance by radiometric calibration; this workflow is illustrated for a large FWF lidar campaign in the Ötztal Alps (Tyrol, Austria), operated with an Optech ALTM 3100 system. The geometric performance of the presented procedure was evaluated by means of a relative and an absolute accuracy assessment using strip differences and orthophotos, resp. The diffuse Lambertian reflectance was evaluated at two rock glaciers within the mentioned lidar campaign. This feature showed good performance for the delineation of the rock glacier boundaries, especially at their lower parts.

Electronic And Structural Study Of Pt-modified Au Vicinal Surfaces: A Model System For Pt-au Catalysts.

Two single crystalline surfaces of Au vicinal to the (111) plane were modified with Pt and studied using scanning tunneling microscopy (STM) and X-ray photoemission spectroscopy (XPS) in ultra-high vacuum environment. The vicinal surfaces studied are Au(332) and Au(887) and different Pt coverage (θPt) were deposited on each surface. From STM images we determine that Pt deposits on both surfaces as nanoislands with heights ranging from 1 ML to 3 ML depending on θPt. On both surfaces the early growth of Pt ad-islands occurs at the lower part of the step edge, with Pt ad-atoms being incorporated into the steps in some cases. XPS results indicate that partial alloying of Pt occurs at the interface at room temperature and at all coverage, as suggested by the negative chemical shift of Pt 4f core line, indicating an upward shift of the d-band center of the alloyed Pt. Also, the existence of a segregated Pt phase especially at higher coverage is detected by XPS. Sample annealing indicates that the temperature rise promotes a further incorporation of Pt atoms into the Au substrate as supported by STM and XPS results. Additionally, the catalytic activity of different PtAu systems reported in the literature for some electrochemical reactions is discussed considering our findings.

Analysis Of The Fluorescence Of Body Fluids On Different Surfaces And Times.

The use of screening techniques, such as an alternative light source (ALS), is important for finding biological evidence at a crime scene. The objective of this study was to evaluate whether biological fluid (blood, semen, saliva, and urine) deposited on different surfaces changes as a function of the age of the sample. Stains were illuminated with a Megamaxx™ ALS System and photographed with a Canon EOS Utility™ camera. Adobe Photoshop™ was utilized to prepare photographs for analysis, and then ImageJ™ was used to record the brightness values of pixels in the images. Data were submitted to analysis of variance using a generalized linear mixed model with two fixed effects (surface and fluid). Time was treated as a random effect (through repeated measures) with a first-order autoregressive covariance structure. Means of significant effects were compared by the Tukey test. The fluorescence of the analyzed biological material varied depending on the age of the sample. Fluorescence was lower when the samples were moist. Fluorescence remained constant when the sample was dry, up to the maximum period analyzed (60 days), independent of the substrate on which the fluid was deposited, showing the novelty of this study. Therefore, the forensic expert can detect biological fluids at the crime scene using an ALS even several days after a crime has occurred.

Influence of implant surfaces on osseointegration

The biological fixation between the dental implant surfaces and jaw bones should be considered a prerequisite for the long-term success of implant-supported prostheses. In this context, the implant surface modifications gained an important and decisive place in implant research over the last years. As the most investigated topic in, it aided the development of enhanced dental treatment modalities and the expansion of dental implant use. Nowadays, a large number of implant types with a great variety of surface properties and other features are commercially available and have to be treated with caution. Although surface modifications have been shown to enhance osseointegration at early implantation times, for example, the clinician should look for research evidence before selecting a dental implant for a specific use. This paper reviews the literature on dental implant surfaces by assessing in vitro and in vivo studies to show the current perspective of implant development. The review comprises quantitative and qualitative results on the analysis of bone-implant interface using micro and nano implant surface topographies. Furthermore, the perspective of incorporating biomimetic molecules (e.g.: peptides and bone morphogenetic proteins) to the implant surface and their effects on bone formation and remodeling around implants are discussed.

Enhanced bone apposition to Brazilian microrough titanium surfaces

It has recently been reported that machined and microrough (micro) Brazilian titanium (Ti) implants have good production standards. The aim of this study was to evaluate in vivo bone formation around 2 different implant surfaces placed in dog's mandible. Thirty-two screw-typed Ti implants were used in this study. Mandibular premolars were extracted in 8 dogs and, after 12 weeks, 2 machined (Neodent Titamax, Brazil) and 2 micro implants (Neodent Titamax Porous, Brazil) were placed in each animal. Biopsies were taken at 3 and 8 weeks post-implantation and stained with Stevenel's blue and Alizarin red for histomorphometric measurements of bone-to-implant contact (BIC), bone area between threads (BABT) and bone area within the mirror area (BAMA). Data were analyzed statistically by two-way ANOVA (α=0.05). While at 3 weeks micro implants exhibited significantly more BIC than machined ones (55 ± 12.5% and 35.6 ± 15%, p<0.05), no significant difference in such parameter was detected at 8 weeks (51.2 ± 21% and 48.6 ± 18.1%, p>0.05). There were no significant differences in BABT and BAMA between the implants. Micro surfaces promoted higher contact osteogenesis. These data indicate that this commercial micro Ti implant surface enhances contact osteogenesis at an early post-implantation period when compared to the machined one.

Characterization of thiol-functionalised silica films deposited on electrode surfaces

Thiol-functionalised silica films were deposited on various electrode surfaces (gold, platinum, glassy carbon) by spin-coating sol-gel mixtures in the presence of a surfactant template. Film formation occurred by evaporation induced self-assembly (EISA) involving the hydrolysis and (co)condensation of silane and organosilane precursors on the electrode surface. The characterization of such material was performed by IR spectroscopy, thermogravimetry (TG), elemental analysis (EA), atomic force microscopy (AFM), scanning electron microscopy (SEM) and cyclic voltammetry (CV).

Shoulder Muscular Activity During Isometric Three-Point Kneeling Exercise on Stable and Unstable Surfaces

The purpose of this study was to determine if performing isometric 3-point kneeling exercises on a Swiss ball influenced the isometric force output and EMG activities of the shoulder muscles when compared with performing the same exercises on a stable base of support. Twenty healthy adults performed the isometric 3-point kneeling exercises with the hand placed either on a stable surface or on a Swiss ball. Surface EMG was recorded from the posterior deltoid, pectoralis major, biceps brachii, triceps brachii, upper trapezius, and serratus anterior muscles using surface differential electrodes. All EMG data were reported as percentages of the average root mean square (RMS) values obtained in maximum voluntary contractions for each muscle studied. The highest load value was obtained during exercise on a stable surface. A significant increase was observed in the activation of glenohumeral muscles during exercises on a Swiss ball. However, there were no differences in EMG activities of the scapulothoracic muscles. These results suggest that exercises performed on unstable surfaces may provide muscular activity levels similar to those performed on stable surfaces, without the need to apply greater external loads to the musculoskeletal system. Therefore, exercises on unstable surfaces may be useful during the process of tissue regeneration.

Nd:YAG Laser Improves Biocompatibility of Human Dental Root Surfaces

Objective: Our goal was to compare the in vivo biocompatibility of dental root surfaces submitted to four different treatments after tooth avulsion followed by implantation into rat subcutaneous tissue. Background Data: Dental root surface preparation prior to replanting teeth remains a challenge for endodontists. Root surface changes made by Nd:YAG irradiation could be an alternative preparation. Methods: Forty-eight freshly extracted human dental roots were randomly divided into four treatment groups prior to implantation into rat subcutaneous tissue: G1, dry root, left in the environment up to 3 h; G2, the same treatment as G1, followed by a soaking treatment in a 2.4% sodium fluoride solution (pH 5.5); G3, root soaked in physiologic saline after avulsion for 72 h; G4, the same treatment as G1, followed by Nd:YAG laser irradiation (2.0 W, 20 Hz, 100 mJ, and 124.34 J/cm(2)). The animals were sacrificed 1, 7, and 45 d later. Histological and scanning electron microscopy analyses were done. Results: All dental roots were involved and in intimate contact with connective tissue capsules of variable thicknesses. Differences were observed in the degree of inflammation and in connective tissue maturation. In G3 the inflammatory infiltrate was maintained for 45 d, whereas the Nd:YAG laser irradiation (G4) led to milder responses. The overall aspects of the root surfaces were similar, except by the irradiated roots, where fusion and resolidification of the root surface covering the dentinal tubules were observed. Conclusion: Nd:YAG laser irradiation improves the biocompatibility of dental root and thus could be an alternative treatment of dental root prior to replantation.

Morphologic changes and removal of debris on apical dentin surfaces after Nd : YAG laser and diode laser irradiation

Objective: To verify the effects of laser energy on intracanal dentin surfaces, by analyzing the morphologic changes and removal of debris in the apical third of 30 extracted human teeth, prepared and irradiated with the Nd:YAG laser and diode laser. Background Data: Lasers have been widely used in endodontics. The morphologic changes in dentin walls caused by Nd: YAG and diode laser irradiation could improve apical seals and cleanliness. Materials and Methods: The protocol used for Nd: YAG laser irradiation was 1.5 W, 100 mJ, and 15 Hz, in pulsed mode, and for diode laser was 2.5 W in continuous mode. Each specimen was irradiated four times at a speed of 2 mm/sec with a 20-sec interval between applications. Five calibrated examiners scored the morphologic changes and debris removal on a 4-point scale. Results: In analyzing the scores, there were no statistically significant differences between the two types of laser for either parameter, according to Kruskal-Wallis testing at p = 0.05. The SEM images showed fusion and resolidification of the dentin surface, with partial removal of debris on the specimens irradiated with the Nd: YAG laser and the diode laser, compared with controls. Conclusion: Both lasers promote morphologic changes and debris removal. These alterations of the dentin surface appeared to be more evident in the Nd: YAG laser group, but the diode laser group showed more uniform changes.

Er : YAG laser, ultrasonic system, and curette produce different profiles on dentine root surfaces: An in vitro study

Objective: The aim of the present study was to compare the in vitro effects of the Er:YAG laser, an ultrasonic system, and manual curette on dentine root surface by roughness and micro-morphological analysis. Materials and Methods: Thirty-six flattened bovine roots were randomly assigned to one of the following groups: group 1 (n = 12): Er: YAG laser ( 2940 nm), 120 mJ/pulse, 10 Hz, 8.4 J/cm(2); group 2 ( n = 12): ultrasonic system; and group 3 ( n = 12): manual curette. The mean surface roughness (Ra) of each sample was measured using a profilometer before and after the treatments. The micro-morphology of the treated and untreated ( control) root surfaces was evaluated with scanning electron microscopy (SEM) at 50 x and 1000 x magnification. Results: Analysis with the profilometer showed that for equal times of instrumentation, the smoothest surfaces were produced by the Er: YAG laser and the ultrasonic system, followed by the curette ( p < 0.05). Morphological analyses demonstrated that treatment with the Er: YAG laser produced some areas with an irregular surface, craters, and ablation of the intertubular dentin. The smear layer was removed and dentine tubules were opened by both curettes and the ultrasonic system. The micro-morphology of the dentine root surface after ultrasonic treatment, however, demonstrated randomly distributed areas cratering. Conclusion: All instruments increased the roughness of the dentine root surface after treatment; however, the curette produced rougher surfaces than the other devices. SEM analysis revealed distinct root surface profiles produced by the three devices.

FLOWS WITHOUT WANDERING POINTS ON COMPACT CONNECTED SURFACES

Given a compact 2 dimensional manifold M we classify all continuous flows phi without wandering points on M. This classification is performed by finding finitely many pairwise disjoint open phi-invariant subsets {U(1), U(2), ..., U(n)} of M such that U(i=1)(n) (U(i)) over bar = M and each U(i) is either a suspension of an interval exchange transformation, or a maximal open cylinder made up of closed trajectories of phi.

Design and fabrication of microcavity-array superhydrophobic surfaces

We have modeled, fabricated, and characterized superhydrophobic surfaces with a morphology formed of periodic microstructures which are cavities. This surface morphology is the inverse of that generally reported in the literature when the surface is formed of pillars or protrusions, and has the advantage that when immersed in water the confined air inside the cavities tends to expel the invading water. This differs from the case of a surface morphology formed of pillars or protrusions, for which water can penetrate irreversibly among the microstructures, necessitating complete drying of the surface in order to again recover its superhydrophobic character. We have developed a theoretical model that allows calculation of the microcavity dimensions needed to obtain superhydrophobic surfaces composed of patterns of such microcavities, and that provides estimates of the advancing and receding contact angle as a function of microcavity parameters. The model predicts that the cavity aspect ratio (depth-to-diameter ratio) can be much less than unity, indicating that the microcavities do not need to be deep in order to obtain a surface with enhanced superhydrophobic character. Specific microcavity patterns have been fabricated in polydimethylsiloxane and characterized by scanning electron microscopy, atomic force microscopy, and contact angle measurements. The measured advancing and receding contact angles are in good agreement with the predictions of the model. (C) 2010 American Institute of Physics. [doi:10.1063/1.3466979]