近紅外線影像由於其物理特性與成像方式，其紋理細節都有發散模糊不清的現象，對於以紋理為主要特徵的年齡辨識問題而言更具挑戰。本論文主要目的是以近紅外線人臉影像為基礎，找出對近紅外線年齡特徵有最佳描述力的特徵描述子，辨識近紅外線影像中被拍攝者的年齡區間，建構整個年齡層估算機制。 相關研究一部分關注可見光年齡辨識，另一部分則聚焦在近紅外線人臉辨識，目前還沒有近紅外線年齡辨識的相關文獻能參考，如何從接近的研究領域找尋是適當的演算法是本研究遇到的第一個挑戰。在資料庫的部分，FGNET和MORPH常被用於可見光的年齡辨識議題； PolyU和LDHF則用於近紅外線人臉辨識相關研究，在目前沒有近紅外線年齡資料庫的情況，本研究自建RSNIR（Intel RealSense Near-Infrared Age Database），因此如何標準化拍攝環境流程、蒐集穩定的近紅外線影像是本研究面臨的第二個挑戰。 區域性特徵擷取方法的關鍵在於特徵描述子的描述力。本研究以LBP（Local Binary Patterns）為基礎，探討LBP在內的24個特徵描述子，最後實驗測試各個描述子在RSNIR的辨識率，結果發現基本型Fuzzy LBP和擴充型RILBP對近紅外線年齡特徵有最佳描述力。在空間譜子區塊（patch）設計部分，以3x3切割子區塊數的辨識效果最好，反應出其與影像校正時的人臉影像空間定義方式有關。

The wrist is a body part that can be used during repetitive movements in many work environments. There is a need to measure these movements in order to notice harmful repetitive movements in advance. There are many different ways to measure these movements, such as with the use of a depth camera. The goal of this study is to determine if this can be done with high precision compared to other technologies. In order to determine this, an application was created that used several different technologies and libraries to track and pinpoint the hand’s and forearm’s location in each frame. With these locations, together with timestamps from the frames, the angular velocity of the wrist could be calculated. The recordings were made in several different test cases with factors such as background, clothes and lighting changing in each test. In order to compare the depth cameras values, a golden standard had to be set. The depth camera’s recorded values were compared to the golden standard’s recorded values by displaying the values on a graph and by calculating the root mean squared error as well as the mean absolute error. The results indicated that a depth camera can be used to measure wrist movements relatively accurately, even with more advanced movements relative to this study. The result also showed that the depth camera had problems in some test cases. / Handleden är en kroppsdel som kan användas under repetitiva rörelser i många arbetsmiljöer. Det finns ett behov av att mäta dessa rörelser för att upptäcka skadliga repetitiva rörelser i förväg. Det finns många olika sätt att mäta dessa rörelser, till exempel med hjälp av en djupkamera. Målet med denna studie är att avgöra om detta kan göras med hög precision jämfört med andra teknologier. För att avgöra detta skapades en applikation som använder flera olika teknologier och bibliotek för att spåra och lokalisera handens och underarmens position i varje bildruta. Med hjälp av dessa positioner, tillsammans med tidsstämplar från bildrutorna, kunde vinkelhastigheten för handleden beräknas. Inspelningarna gjordes i flera olika testfall där faktorer som bakgrund, kläder och belysning ändrades i varje test. För att kunna jämföra djupkamerans värden behövdes en referensstandard fastställas. Djupkamerans inspelade värden jämfördes med referensstandardens inspelade värden genom att visa värdena på en graf och beräkna rotmedelkvadratfelet samt medelabsolutfelet. Resultaten indikerade att en djupkamera kan användas för att mäta handledsrörelser relativt noggrant, även med mer avancerade rörelser i förhållande till denna studie. Resultatet visade även att djupkameran hade problem i vissa testfall.

This thesis deals with construction of a cheap robotic manipulator, which should be used for exhibitions and educational purposes. This project is a teamwork of two students. A robotic arm with four degrees of freedom was developed. Control and power electronics were installed for whole robot. The software’s aim is to develop a software that can control the robot by visual feedback, obtained from image processing of an image from 3D camera. Lastly, a graphic user interface for robot movement control is presented.

The aim of this master’s thesis is to describe the Robotic Vision for Bin Picking usage and creating an application for the realization of this task. This application will be able to distinguish several objects based on data from a camera with deep perception and should find the location of object, recognize it and determine its location and orientation. Bin Picking is one of the biggest challenges in today's automation.

<p> This thesis presents work and simulations containing the use of Artificial Intelligence for real-time perception and real-time anomaly detection using the computer and sensors onboard an Unmanned Aerial Vehicle. One goal of this research is to develop a highly accurate, high-performance computer vision system that can then be used as a framework for object detection, obstacle avoidance, motion estimation, 3D reconstruction, and vision-based GPS denied path planning. The method developed and presented in this paper integrates software and hardware techniques to reach optimal performance for real-time operations. </p> <p>This thesis also presents a solution to real-time anomaly detection using neural networks to further the safety and reliability of operations for the UAV. Real-time telemetry data from different sensors are used to predict failures before they occur. Both these systems together form the framework behind the Intelligent UAV platform, which can be rapidly adopted for different varieties of use cases because of its modular nature and on-board suite of sensors. </p>