Orchard fruits and vegetable crops are vulnerable to wild birds and animals. These wild birds and animals can cause critical damage to the produce. Traditional methods of scaring away birds such as scarecrows are not long-term solutions but short-term solutions. This is a huge problem especially near areas like San Luis Obispo where there are vineyards. Bird damage can be as high as 50% for grapes being grown in vineyards. The total estimated revenue lost annually in the 10 counties in California due to bird and rodent damage to 22 selected crops ranged from $168 million to $504 million (in 2009 dollars). A more effective and permanent system needs to be put into place. Monitoring systems in agricultural settings could potentially provide a lot of data for image processing. Most current monitoring systems however don’t focus on image processing but instead really heavily on sensors. Just having sensors for certain systems work, but for birds, monitoring it is not an option because they are not domesticated like pigs, cows etc. in which most these agricultural monitoring systems work on. Birds can fly in and out of the area whereas domesticated animals can be confined to certain physical regions. The most crucial step in a smart scarecrow system would be how a threat would v be detected. Image processing methods can be effectively applied to detecting items in video footage. This paper will focus on bird detection and will analyze motion detection with image subtraction, bird detection with template matching, and bird detection with the Viola-Jones Algorithm. Of the methods considered, bird detection with the Viola-Jones Algorithm had the highest accuracy (87%) with a somewhat low false positive rate. This image processing step would ideally be incorporated with hardware (such as a microcontroller or FPGA, sensors, a camera etc.) to form a smart scarecrow system.

The goal of this diploma thesis is to design and implement a mobile application for home security system on the iOS platform. The application introduces a complex solution allowing the transmission of the audio and video streams between the paired mobile devices using WebRTC. The final module represents universal solution for peer-to-peer audio and video communication. The thesis also deals with the field of computer vision, namely efficient motion detection algorithms. The module for motion detection implements ViBe algorithm using Metal. In case that the motion is detected the application notifies a user with a push notification. Synchronization of application data is implemented using Cloudkit and the data persistance using Realm library.

The aim of this diploma thesis is to obtain information on the motion of people in a scene from the record of the stationary camera. The procedure to detect exceptional events in the scene was designed. Exceptional events can be fast-moving persons, or persons moving in di erent places than everyone else in the scene. To trace the motion of persons, two algorithms were applied and tested - Optical flow and CAMSHIFT. The analysis of the resulting motions is performed by monitoring the progress of motion, and its comparison with the other motions in the scene. The analysis result is represented by detected exceptional motions that can be found in the video. The areas where the motion occurs in the scene, and where the motion is the most common are also described together with the motion direction analysis. The exceptional motion parts extracted from the video represent the main result of the work.

This master's thesis deals with fire detection in videosequences. Attention is paid to the known characteristics of fire and basic principles of existing solutions which deal with this issue. The thesis also describes design, implementation and testing of a fire detector that is based on the recognition of suspicious areas by fire color modeling, combined with detection of motion and light intensity variations.

Shortages of electricity in larger cities in Sweden are becoming more common and might increase with an increased consumption. The thesis focuses on lights due to approximately 19 - 20% of the total electricity consumption coming from lights. It examines whether a control system consisting of an Motion Detection Dimming (MDD) System combined with a Daylight-Linked Control System (DLCS) that illuminates the workspace of individual employees while avoiding bright spots in a Swedish office context achieves an energy gain. An MDD system activates different lamps depending on if motion was detected while the DLCS adjusts the brightness the lights emit depending on the natural light in the area. Data was gathered which connected spans of cloudiness to lux (lx) values. This allowed for simulations for the year with the least amount of brightness and the year with the most amount of brightness, since the 2000s. Workers movement in an office environment was also documented. The results for both of the data gatherings where imported in to a simulation that calculated the electrical energy, in kWh. Three scenarios were created and simulated; all workers being at the office moving around, hybrid working and lastly all employees being present but never leaving their desks. The results showed that an energy gain could be achieved from all three scenarios, 48,9 - 50,1% for the first, 40,1 - 41,2% for the second and 4,4 - 5,5% for the final, compared to the offices current consumption. According to the results an MDD-system would lower the energy consumption, in a Swedish office environment, without creating a ”bright spot”. This meant that the master thesis succeeded in proving that it was, with high probability, possible. However, the savings would not increase significantly with the Daylight-Linked Control System. / Elektricitetsbrist i storstäder i Sverige blir mer vanliga och kan öka med en ökad elkonsumption. Arbetet fokuserar på ljus eftersom cirka 19 - 20% av den totala energikonsumptionen kommer från ljus. Projektet undersöker om ett Motion Detection Dimming (MDD) System kombinerat med ett Daylight-Linked Control System (DLCS) som lyser upp arbetsområdet för individuella anställda medan ljusa fläckar undviks i ett Svenskt kontorssammanhang kan spara energi. Ett MDD system aktiverar olika lampor beroende på om rörelse blev detekterat medan ett DLCS justerar styrkan på lamporna beroende på naturliga ljuset i omgivningen. Data som kopplade ihop molnighets-spann med lux-värden samlades ihop. Detta möjliggjorde simuleringar för ljusaste och mörkaste året efter år 2000 att göras. Hur anställda rörde sig i kontorsmiljön dokumenterades även. Resultaten från datainsamlingarna importerades in till en simulering som beräknade elenergin, i kWh. Simuleringarna utfördes för tre scenarios; alla arbetare är på kontoret och rör sig i det designerade området, hybridarbete och slutligen alla anställda är på kontoret men lämnar aldrig sina skrivbord. Resultatet påvisade att en energivinst kunde uppnås i alla tre scenarios, 48,9 - 50,1% för det första, 40,1 - 41,2% för den andra och 4,4 - 5,5% för det sista, jämfört med kontorets nuvarandra energikonsumption. Enligt resultaten minskar ett MDD-system energikonsumptionen,i en svensk kontorsmiljö, utan att bilda ljusfläckar. Detta medför att arbetet lyckades bevisa att det är, med hög sannolikhet möjligt. Å andra sidan visade det sig att besparingarna för Daylight-Linked Control System inte var stora i jämförelse.

One of the major goals of this thesis is to investigate the extent to which correspondence noise, (i.e., the false pairing of dots in adjacent frames) limits motion detection performance in random dot kinematograms (RDKs). The performance measures of interest are Dmax and Dmin i.e., the largest and smallest inter-frame dot displacement, respectively, for which motion can be reliably detected. Dmax and threshold coherence (i.e., the smallest proportion of dots that must be moved between frames for motion to be reliably detected) in RDKs are known to be affected by false pairing or correspondence noise. Here the roles of correspondence noise and receptive field geometry in limiting performance are investigated. The range of Dmax observed in the literature is consistent with the current information-limit based interpretation. Dmin is interpreted in the light of correspondence noise and under-sampling. Based on the psychophysical experiments performed in the early parts of the dissertation, a model for correspondence noise based on the principle of receptive field scaling is developed for Dmax. Model simulations provide a good account of psychophysically estimated Dmax over a range of stimulus parameters, showing that correspondence noise and receptive field geometry have a major influence on displacement thresholds.

With the growth of computer power, Digital Image Processing plays a more and more important role in the modern world, including the field of industry, medical, communications, spaceflight technology etc. There is no clear definition how to divide the digital image processing, but normally, digital image processing includes three main steps: low-level, mid-level and highlevel processing. Low-level processing involves primitive operations, such as: image preprocessing to reduce the noise, contrast enhancement, and image sharpening. Mid-level processing on images involves tasks such as segmentation (partitioning an image into regions or objects), description of those objects to reduce them to a form suitable for computer processing, and classification (recognition) of individual objects. Finally, higher-level processing involves "making sense" of an ensemble of recognised objects, as in image analysis. Based on the theory just described in the last paragraph, this thesis is organised in three parts: Colour Edge and Face Detection; Hand motion detection; Hand Gesture Detection and Medical Image Processing. II In Colour Edge Detection, two new images G-image and R-image are built through colour space transform, after that, the two edges extracted from G-image and R-image respectively are combined to obtain the final new edge. In Face Detection, a skin model is built first, then the boundary condition of this skin model can be extracted to cover almost all of the skin pixels. After skin detection, the knowledge about size, size ratio, locations of ears and mouth is used to recognise the face in the skin regions. In Hand Motion Detection, frame differe is compared with an automatically chosen threshold in order to identify the moving object. For some special situations, with slow or smooth object motion, the background modelling and frame differencing are combined in order to improve the performance. In Hand Gesture Recognition, 3 features of every testing image are input to Gaussian Mixture Model (GMM), and then the Expectation Maximization algorithm (EM)is used to compare the GMM from testing images and GMM from training images in order to classify the results. In Medical Image Processing (mammograms), the Artificial Neural Network (ANN) and clustering rule are applied to choose the feature. Two classifier, ANN and Support Vector Machine (SVM), have been applied to classify the results, in this processing, the balance learning theory and optimized decision has been developed are applied to improve the performance.

This thesis focuses on visual multiple-target tracking (MTT) from a UAV. Typical state-of-the-art multiple-target trackers rely on an object detector as the primary detection source. However, object detectors usually require a GPU to process images in real-time, which may not be feasible to carry on-board a UAV. Additionally, they often do not produce consistent detections for small objects typical of UAV imagery.In our method, we instead detect motion to identify objects of interest in the scene. We detect motion at corners in the image using optical flow. We also track points long-term to continue tracking stopped objects. Since our motion detection algorithm generates multiple detections at each time-step, we use a hybrid probabilistic data association filter combined with a single iteration of expectation maximization to improve tracking accuracy.We also present a motion detection algorithm that accounts for parallax in non-planar UAV imagery. We use the essential matrix to distinguish between true object motion and apparent object motion due to parallax. Instead of calculating the essential matrix directly, which can be time-consuming, we design a new algorithm that optimizes the rotation and translation between frames. This new algorithm requires only 4 ms instead of 47 ms per frame of the video sequence.We demonstrate the performance of these algorithms on video data. These algorithms are shown to improve tracking accuracy, reliability, and speed. All these contributions are capable of running in real-time without a GPU.

<p>Neregularnost prostiranja radio talasa je uobičajeni fenomen koji<br />utiče na kvalitet radio veze u okviru bežične mreže, rezultujući<br />različitim obrascima prostiranja radio talasa. Ova teza daje<br />predlog nekoliko postupaka analize prostiranja radio talasa u cilju<br />bez-senzorskog otkrivanja prisustva i kretanja ljudi unutar postojeće<br />bežične mreže. Indikator primljene snage radio signala predstavlja<br />osnovni element analize, iz kog se izdvajaju informaciono,<br />amplitudsko i frekventno obeležje. Analizom navedenih obeležja<br />moguća je realizacija robusnog postupka bez-senzorske detekcije ljudi<br />koja se može primeniti u različitim rešenjima ambijentalne<br />inteligencije, zahtevajući minimalan broj elemenata fizičke<br />arhitekture, neophodnih za uspostavljanje korisnički svesnog<br />okruženja.</p> / <p>Radio irregularity is a common and non-negligible phenomenon that impacts<br />the connectivity and interference in a wireless network, by introducing<br />disturbances in radio signal&rsquo;s propagation pattern. In order to detect a<br />possible presence of a human subject within the existing radio network<br />sensorlessly, this thesis analyze the irregularity data expressed in a form of<br />received signal strength variation. The received signal strength variation is<br />decomposed into information, amplitude and frequency characteristics. The<br />combination of these three characteristics analysis enables the definition of<br />robust and cost-effective device-free human presence detection method that<br />can be exploited for various ambient intelligence solutions, requiring the<br />minimum hardware add-ons that are necessary for the establishment of a<br />user aware environment.</p>

<p>Mотив за истраживање је унапређење процеса аутоматског препознавања говорника без обзира на садржај изговоренoг текста.<br />Циљ ове докторске дисертације је дефинисање новог биометријског обележја за препознавање говорника независно од изговореног текста &minus; дводимензионалне информационе ентропије говорног сигнала.<br />Дефинисање новог обележја се врши искључиво у временском домену, па је рачунарска сложеност алгоритма за његово издвајање знатно мања у односу на обележја која се издвајају у фреквенцијском домену. Оцена перформанси дводимензионалне информационе ентропије је урађена над репрезентативним скупом случајно одабраних говорника. Показано је да предложено обележје има малу варијабилност унутар говорног сигнала једног говорника, а велику варијабилност између говорних сигнала различитих говорника.</p> / <p>Motiv za istraživanje je unapređenje procesa automatskog prepoznavanja govornika bez obzira na sadržaj izgovorenog teksta.<br />Cilj ove doktorske disertacije je definisanje novog biometrijskog obeležja za prepoznavanje govornika nezavisno od izgovorenog teksta &minus; dvodimenzionalne informacione entropije govornog signala.<br />Definisanje novog obeležja se vrši isključivo u vremenskom domenu, pa je računarska složenost algoritma za njegovo izdvajanje znatno manja u odnosu na obeležja koja se izdvajaju u frekvencijskom domenu. Ocena performansi dvodimenzionalne informacione entropije je urađena nad reprezentativnim skupom slučajno odabranih govornika. Pokazano je da predloženo obeležje ima malu varijabilnost unutar govornog signala jednog govornika, a veliku varijabilnost između govornih signala različitih govornika.</p> / <p>Тhe motivation for the research is the improvement of the automatic speaker recognition process regardless of the content of spoken text.<br />The objective of this dissertation is to define a new biometric text-independent speaker recognition feature &minus; the two-dimensional informational entropy of speech signal.<br />Definition of the new feature is performed in time domain exclusively, so the computing complexity of the algorithm for feature extraction is significantly lower in comparison to feature extraction in spectral domain. Performance analysis of two-dimensional information entropy is performed on the representative set of randomly chosen speakers. It has been shown that new feature has small within-speaker variability and significant between-speaker variability.</p>