Search

In-cloud icing of objects is caused by supercooled microscopic water droplets carried by the wind. To estimate the icing rate of objects in such conditions, the liquid water content (LWC) of the icing cloud and the median volume diameter (MVD) of the droplets are measured. Mixed-phase clouds also contain ice crystals that must be ruled out in order to avoid the overestimation of the icing rate. Typically, cloud droplet instruments are not able to do this. A particle imaging instrument icing condition evaluation method (ICEMET) was used to observe in-cloud icing conditions. This lensless device uses a computational imaging method to reconstruct the shadow images of the microscopic objects. The size, position, and shape descriptors of each particle are measured. These data are then used to filter out the ice crystals. The droplet size distribution and the size of the measurement volume are used to determine the LWC and MVD. The performance of the instrument was tested under mixed-phase icing conditions in a wind tunnel and on a wind turbine. The measured LWC and MVD values were used to model the ice accretion on a cylinder-shaped object according to the ISO 12494:2017 icing standard. In the wind tunnel, the modeled ice mass was compared with the weighed ice mass collected by a cylinder. According to our results, ice accretion rates were overestimated by 65.6% on average without filtering out the ice crystals. Thus, the ability to distinguish between droplets and ice crystals is essential for estimating the icing rate properly.

Objective: The aim of this study was to investigate the reliability of digital imaging for detecting restorative treatment need among individuals in their 20s by comparing the outcome of digital imaging with clinical caries findings at the patient level. Material and methods: Five intraoral clinical daylight and digital fluorescence images were taken extraorally of 21 patients. A clinical examination was then performed by a trained and calibrated dentist. Additionally, the patients answered a multiple-choice questionnaire about their health habits. The images were analysed and caries findings were recorded. For statistical analysis, sensitivity and specificity were calculated. Results were shown as ROC curves and AUC values. All analyses were done using SPSS (version 24.0, Chicago, IL). Results: Caries lesions were most often detected in molars and least often in canines. When using the clinical status as gold standard, digital imaging gave an AUC value of 0.617, whereas the outcome by questionnaire gave an AUC value of 0.719. When using the combined outcome of digital imaging and the questionnaire, the AUC value was 0.694 with clinical validation. Conclusions: It can be concluded that health kiosks may help to reduce the number of patients waiting for dental treatment; more specifically, the questionnaire with individual feedback may provide a new instrument for providing instructions for homecare online. However, the camera system must be developed further, and dentists and dental hygienists require training to analyse the images.

An optical cloud droplet and ice crystal measurement system ICEMET (icing condition evaluation method), designed for present icing condition monitoring in field conditions, is presented. The aim in this work has been to develop a simple but precise imaging technique to measure the two often missing parameters needed in icing rate calculations caused by icing clouds—the droplet size distribution (DSD) and the liquid water content (LWC) of the air. The measurement principle of the sensor is based on lens-less digital in-line holographic imaging. Cloud droplets and ice crystals are illuminated by a short laser light pulse and the resulting hologram is digitally sampled by a digital image sensor and the digital hologram is then numerically analyzed to calculate the present DSD and LWC values. The sensor has anti-icing heating power up to 500 W and it is freely rotating by the wind for an optimal sampling direction and aerodynamics. A volume of 0.5 cm³ is sampled in each hologram and the maximum sampling rate is 3 cm³/s. Laboratory tests and simulations were made to ensure the adequate operation of the measurement sensor. Computational flow dynamics simulations showed good agreement with droplet concentration distributions measured from an icing wind tunnel. The anti-icing heating of the sensor kept the sensor operational even in severe icing conditions; the most severe test conditions were the temperature − 15 °C, wind speed 20 m/s and the LWC 0.185 g/m³. The verification measurements made using NIST traceable monodisperse particle standard glass spheres showed that the ICEMET sensor measurement median diameter 25.54 µm matched well with 25.60 µm ± 0.70 µm diameter confidence level given by the manufacturer.

Field measurement results from a novel optical cloud droplet monitoring sensor designed for icing conditions monitoring are presented. The sensor has been demonstrated at two sites in northern Finland; first at Global Atmosphere Watch Station in Pallas together with a reference icing sensor and secondly mounted on a wind turbine nacelle in eastern Finland in 2017. Test runs in an icing wind tunnel have been made where more severe icing conditions were generated. The ICEMET sensor measurement principle is based on capturing the images of cloud droplets and ice particles. Droplet properties, such as droplet size distribution (DSD) and median volume diameter (MVD), are acquired by means of image analysis of the captured images. The images and the calculated features (size, location, shape descriptors) of all the found particles are saved in a database. A volume of 0.5 cm³ is imaged in a single frame. The liquid water content (LWC) is calculated based on this known sample volume in combination with the droplet data acquired from the image analysis of the found and filtered particles (droplets only). The sensor is typically freely rotating — it aligns itself against the wind by a wing on the backside. In the rotating configuration, the maximum sampling rate is 3 cm³/s. The movement of the particles inside sample volume is frozen in the images by a nanosecond scale light flash, making the sample volume independent of the wind speed. The maximum wind speed tested in a wind tunnel with the sensor is 40 m/s. The cloud droplet sizes from 5 to 200 microns are measured by the ICEMET sensor. In this paper LWC and MVD measurement results from the field tests and the wind tunnel tests with the sensor are presented and discussed. The webpages for the sensor can be found at https://www.oulu.fi/icemet.

Holografia on kolmiulotteisia kuvia eli hologrammeja käsittelevä tiede. Vastaavasti digitaalisella holografialla tarkoitetaan holografisten menetelmien suorittamista digitaalisten laitteiden avulla. Tämä sisältää hologrammien tallentamisen, synteesin ja rekonstruktion digitaalisten sensorien ja tietokoneiden avulla. Tieteenalana holografia on jo erittäin varttunut ja sillä on nykyään sovelluksia useilla eri aloilla. Eräitä tavallisimpia ovat digitaaliseen holografiaan perustuvat mikroskoopit. Uudempia ja kenties tulevaisuudessa yleistyviä sovelluksia ovat ilmakehän hiukkasia kuvaavat holografiset laitteet. Oulun yliopistossa kehitetty ICEMET-järjestelmä on esimerkki tällaisista meteorologisista sovelluksista. ICEMET:n on suunniteltu käytettäväksi todellisen maailman käyttöolosuhteissa, kuten esimerkiksi tuulivoimaloiden päällä. Vaihtelevien ja usein ankarien käyttöolosuhteidensa vuoksi ICEMET-sensorin ottamissa hologrammikuvissa on paljon taustakohinaa ja häiriötekijöitä. Tämän vuoksi hologrammeja on esikäsiteltävä ennen näiden rekonstruktioprosessia. Tutkielmaa varten kehitettiin ICEMET-järjestelmän osana toimiva digitaalisia hologrammeja esikäsittelevä tietokoneohjelma, Holofast. Ohjelma käsittelee kuvia näytönohjaimen avulla OpenCL-kirjastoa käyttäen. Näytönohjainta hyödyntämällä saavutettiin huomattavia parannuksia suoritusajoissa. Holofast suorittaa hologrammikuville taustanvähennyksen mediaanilla jakamalla ja pyrkii karsimaan tyhjät pisaradataa sisältämättömät kuvat ennen rekonstruktioprosessia. Tyhjän kuvan tunnistukseen kehitettiin dynaamista datan keruuta käyttävä järjestelmä, jossa kuvista kerättiin referenssipinoon asetettavia pistearvoja. Arvot laskettiin tutkimuksessa kehitetyillä tunnistusmenetelmillä. Menetelmät ovat minimi- ja maksimivärisävyjen erotus, varianssi, keskihajonnan ja värisävyjen erotuksen tulo, värisävyjen määrä, Shannonin entropia ja gradienttikuvan keskiarvo. Oikeilla parametreilla käytettynä voidaan tyhjistä kuvista noin puolet karsia. Mikäli pieniä määriä pisaradataa ollaan valmiita menettämään, voidaan jopa 90 % tyhjistä kuvista tunnistaa. Holography is a science that deals with three-dimensional images, holograms. Digital holography means performing holographic methods using digital devices. This includes saving, synthesizing and reconstruction of holograms using digital sensors and computers. Holography is very mature science and it has many applications across all scientific fields. Digital holographic microscopes are one of the more common examples. Holographic devices that take pictures of airborne particles are a much newer application and might become more widespread in the future. ICEMET system, developed in University of Oulu, is a one example of these meteorological applications. ICEMET is designed to be used in real-world conditions such as on top of wind turbines. Because of the varying and often rough operating environment, the digital holograms of ICEMET-sensor contain high amount of interference. This is why the holograms must be preprocessed before reconstruction process. A digital hologram preprocessing software called Holofast was developed for ICEMET. The software processes images with graphics card using OpenCL library. This significantly increased performance. Holofast performs background subtraction using division with median to hologram images and tries to prune blank images before reconstruction process. A dynamic blank image detection system that collects data from images and pushes score values to stack was developed. Scores were calculated using developed detection methods. The methods are maximum difference of color values, variance, standard deviation multiplied by color value difference, number of colors, Shannon entropy and the mean of gradient. With the right parameters almost half of the blank images could be pruned. The prune percentage can be raised up to 90 % with sacrifice of some droplet images.

Tiivistelmä. Todellisuutta laajentavat virtuaalilasit ja datamerkit rakennusten seinillä kuulostavat tieteiselokuvien tulevaisuuden maalailulta, mutta tosiasiassa niihin vaadittava tekniikka on jo olemassa. Tätä uudenlaista informaation esitystapaa, jossa ympäristöön tuodaan virtuaalista sisältöä tietokoneen avulla, kutsutaan lisätyksi todellisuudeksi. Sisältö voi olla esimerkiksi kuvia, ääntä tai jopa hajuja. Merkkipohjainen tunnistus on eräs tapa toteuttaa lisätty todellisuus. Siinä ympäristöön sijoitellaan erilaisia fyysisiä merkkejä, joiden sijainnin ja datasisällön avulla kone saa tarvittavat tiedot sisällön esittämiseen. Käyttäjien ja kehittäjien näkökulmasta merkkipohjainen tunnistus on suhteellisen yksinkertainen tekniikka, mutta tarjoaa samalla paljon erilaisia keinoja todellisuuden lisäämiseen. Tämän lisäksi tarvittavat ohjelmistot ovat jo olemassa; valmiita merkkipohjaista tunnistusta tukevia konenäkökirjastoja on useita. 2000-luvun nopea älykkäiden mobiililaitteiden yleistymien voi mahdollistaa lisätyn todellisuuden tuomisen tavallisten ihmisten arkeen. Sovellusten suorituskykyvaatimukset ovat kuitenkin eräs avainkysymys, johon ei ole vielä selkeää vastausta. Tutkimusta varten kehitettiin merkkipohjaista tunnistusta hyödyntävä lisätyn todellisuuden sovellus C++-kielellä OpenCV-konenäkökirjastoa ja tämän ArUco-moduulia hyödyntäen. Sovellus muokkaa todellisuutta piirtämällä kuvia ympäristöön sijoitettujen vertailumerkkien päälle. Koska laskentatehoiltaan rajalliset mobiililaitteet ovat tällä hetkellä lisätyn todellisuuden yleisin käyttöalusta, suorituskykymittaukset tehtiin pienellä yhden piirilevyn Raspberry Pi -tietokoneella. Mittauksissa selvisi, että lisätyn todellisuuden sovelluksia on mahdollista ajaa varsin heikkotehoisellakin laitteistolla reaaliajassa. OpenCV tarjoaa monipuoliset työkalut sovelluskehitykseen ja merkkipohjainen tunnistus voidaan toteuttaa sen avulla vaivattomasti. Parasta mahdollista suorituskykyä tavoiteltaessa on kuitenkin syytä kokeilla myös vaihtoehtoisia ratkaisuja. OpenCV:tä käytettäessä yhtäaikaisten tunnistettavien merkkien määrä vaikuttaa oleellisesti tunnistusaikoihin. Tähän onkin syytä kiinnittää huomiota lisätyn todellisuuden järjestelmiä suunniteltaessa.Marker-based tracking in augmented reality. Abstract. Virtual glasses that expand reality and data markers on the walls of buildings sound like a future vision from science fiction movies. However, in reality the required technology already exists. This novel way to present information, where virtual content is brought to environment, is called augmented reality. The content can be for example images, audio or even smells. Marker-based tracking is a one way to implement augmented reality. When using it, different kinds of physical markers are placed to the environment. The position and data of these markers provides computer with the necessary information to present the content. From the users’ and developers’ point of view, marker-based tracking is a quite simple technique, but at the same time, it provides many ways to augment reality. In addition to this, the necessary software already exists. There are various complete computer vision libraries that support marker-based tracking. The rapid spread of smart mobile devices in the beginning of the 21st century enables bringing augmented reality to everyday life of average people. However, the system requirements of these kind of software are still a question mark. An application, which makes use of marker-based tracking, was developed for the research using C++ with OpenCV library and its ArUco-module. The application modifies the reality by drawing images on the fiducial markers that are placed in the environment. Because mobile devices with limited performance are currently the most used platform of augmented reality, the performance measurements were taken on a small single-board Rasperry Pi computer. The measurements revealed that it’s possible to run augmented reality applications in real time using fairly weak hardware. OpenCV offers versatile tools for software development and marker-based tracking can be implemented painlessly using it. However, when seeking the best possible performance, it’s worthwhile to also try out alternative solutions. When using OpenCV the number of markers detected simultaneously affects detection times. This should be taken into account when designing augmented reality systems.