Urban Detection From Hyperspectral Images Using Dimension-Reduction Model and Fusion of Multiple Segmentations Based on Stuctural and Textural Features

Ce mémoire de maîtrise présente une nouvelle approche non supervisée pour détecter et segmenter les régions urbaines dans les images hyperspectrales. La méthode proposée n ́ecessite trois étapes. Tout d’abord, afin de réduire le coût calculatoire de notre algorithme, une image couleur du contenu spectral est estimée. A cette fin, une étape de réduction de dimensionalité non-linéaire, basée sur deux critères complémentaires mais contradictoires de bonne visualisation; à savoir la précision et le contraste, est réalisée pour l’affichage couleur de chaque image hyperspectrale. Ensuite, pour discriminer les régions urbaines des régions non urbaines, la seconde étape consiste à extraire quelques caractéristiques discriminantes (et complémentaires) sur cette image hyperspectrale couleur. A cette fin, nous avons extrait une série de paramètres discriminants pour décrire les caractéristiques d’une zone urbaine, principalement composée d’objets manufacturés de formes simples g ́eométriques et régulières. Nous avons utilisé des caractéristiques texturales basées sur les niveaux de gris, la magnitude du gradient ou des paramètres issus de la matrice de co-occurrence combinés avec des caractéristiques structurelles basées sur l’orientation locale du gradient de l’image et la détection locale de segments de droites. Afin de réduire encore la complexité de calcul de notre approche et éviter le problème de la ”malédiction de la dimensionnalité” quand on décide de regrouper des données de dimensions élevées, nous avons décidé de classifier individuellement, dans la dernière étape, chaque caractéristique texturale ou structurelle avec une simple procédure de K-moyennes et ensuite de combiner ces segmentations grossières, obtenues à faible coût, avec un modèle efficace de fusion de cartes de segmentations. Les expérimentations données dans ce rapport montrent que cette stratégie est efficace visuellement et se compare favorablement aux autres méthodes de détection et segmentation de zones urbaines à partir d’images hyperspectrales.

A new label fusion method using graph cuts: application to hippocampus segmentation

The aim of this paper is to develop a probabilistic modeling framework for the segmentation of structures of interest from a collection of atlases. Given a subset of registered atlases into the target image for a particular Region of Interest (ROI), a statistical model of appearance and shape is computed for fusing the labels. Segmentations are obtained by minimizing an energy function associated with the proposed model, using a graph-cut technique. We test different label fusion methods on publicly available MR images of human brains.

Effect of bone graft type on fusion rates following endoscopic anterior scoliosis correction

Bone graft is generally considered fundamental in achieving solid fusion in scoliosis correction and pseudarthrosis following instrumentation may predispose to implant failure. In endoscopic anterior-instrumented scoliosis surgery, autologous rib or iliac crest graft has been utilised traditionally but both techniques increase operative duration and cause donor site morbidity. Allograft bone and bone- morphogenetic-protein alternatives may improve fusion rates but this remains controversial. This study's objective was to compare two-year postoperative fusion rates in a series of patients who underwent endoscopic anterior instrumentation for thoracic scoliosis utilising various bone graft types. Significantly better rates of fusion occurred in endoscopic anterior instrumented scoliosis correction using femoral allograft compared to autologous rib-heads and iliac crest graft. This may be partly explained by the difficulty obtaining sufficient quantities of autologous graft. Lower fusion rates in the autologous graft group appeared to predispose to rod fracture although the clinical consequence of implant failure is uncertain.

Improved maritime target tracker using colour fusion

Searching for humans lost in vast stretches of ocean has always been a difficult task. This paper investigates a machine vision system that addresses this problem by exploiting the useful properties of alternate colour spaces. In particular, the paper investigates the fusion of colour information from the HSV, RGB, YCbCr and YIQ colour spaces within the emission matrix of a Hidden Markov Model tracker to enhance video based maritime target detection. The system has shown promising results. The paper also identifies challenges still needing to be met.

The effect of bone graft type on fusion rates following anterior thoracoscopic scoliosis correction

Bone graft is generally considered fundamental in achieving solid fusion in scoliosis correction and pseudarthrosis following instrumentation may predispose to implant failure. In thoracoscopic anterior-instrumented scoliosis surgery, autologous rib or iliac crest graft has been utilised traditionally but both techniques increase operative duration and cause donor site morbidity. Allograft bone and bone morphogenetic protein (BMP) alternatives may improve fusion rates but this remains controversial. This study's objective was to compare two-year postoperative fusion rates in a series of patients who underwent thoracoscopic anterior instrumentation for thoracic scoliosis utilising various bone graft types.

Fusion of in-vehicle sensor data to develop Intelligent Driver Training System (IDTS)

The over represented number of novice drivers involved in crashes is alarming. Driver training is one of the interventions aimed at mitigating the number of crashes that involve young drivers. To our knowledge, Advanced Driver Assistance Systems (ADAS) have never been comprehensively used in designing an intelligent driver training system. Currently, there is a need to develop and evaluate ADAS that could assess driving competencies. The aim is to develop an unsupervised system called Intelligent Driver Training System (IDTS) that analyzes crash risks in a given driving situation. In order to design a comprehensive IDTS, data is collected from the Driver, Vehicle and Environment (DVE), synchronized and analyzed. The first implementation phase of this intelligent driver training system deals with synchronizing multiple variables acquired from DVE. RTMaps is used to collect and synchronize data like GPS, vehicle dynamics and driver head movement. After the data synchronization, maneuvers are segmented out as right turn, left turn and overtake. Each maneuver is composed of several individual tasks that are necessary to be performed in a sequential manner. This paper focuses on turn maneuvers. Some of the tasks required in the analysis of ‘turn’ maneuver are: detect the start and end of the turn, detect the indicator status change, check if the indicator was turned on within a safe distance and check the lane keeping during the turn maneuver. This paper proposes a fusion and analysis of heterogeneous data, mainly involved in driving, to determine the risk factor of particular maneuvers within the drive. It also explains the segmentation and risk analysis of the turn maneuver in a drive.

Biological performance of a polycaprolactone-based scaffold used as fusion cage device in a large animal model of spinal reconstructive surgery

A bioactive and bioresorbable scaffold fabricated from medical grade poly (epsilon-caprolactone) and incorporating 20% beta-tricalcium phosphate (mPCL–TCP) was recently developed for bone regeneration at load bearing sites. In the present study, we aimed to evaluate bone ingrowth into mPCL–TCP in a large animal model of lumbar interbody fusion. Six pigs underwent a 2-level (L3/4; L5/6) anterior lumbar interbody fusion (ALIF) implanted with mPCL–TCP þ 0.6 mg rhBMP-2 as treatment group while four other pigs implanted with autogenous bone graft served as control. Computed tomographic scanning and histology revealed complete defect bridging in all (100%) specimen from the treatment group as early as 3 months. Histological evidence of continuing bone remodeling and maturation was observed at 6 months. In the control group, only partial bridging was observed at 3 months and only 50% of segments in this group showed complete defect bridging at 6 months. Furthermore, 25% of segments in the control group showed evidence of graft fracture, resorption and pseudoarthrosis. In contrast, no evidence of graft fractures, pseudoarthrosis or foreign body reaction was observed in the treatment group. These results reveal that mPCL–TCP scaffolds could act as bone graft substitutes by providing a suitable environment for bone regeneration in a dynamic load bearing setting such as in a porcine model of interbody spine fusion.

Multi-spectral fusion for surveillance systems

Surveillance systems such as object tracking and abandoned object detection systems typically rely on a single modality of colour video for their input. These systems work well in controlled conditions but often fail when low lighting, shadowing, smoke, dust or unstable backgrounds are present, or when the objects of interest are a similar colour to the background. Thermal images are not affected by lighting changes or shadowing, and are not overtly affected by smoke, dust or unstable backgrounds. However, thermal images lack colour information which makes distinguishing between different people or objects of interest within the same scene difficult. ----- By using modalities from both the visible and thermal infrared spectra, we are able to obtain more information from a scene and overcome the problems associated with using either modality individually. We evaluate four approaches for fusing visual and thermal images for use in a person tracking system (two early fusion methods, one mid fusion and one late fusion method), in order to determine the most appropriate method for fusing multiple modalities. We also evaluate two of these approaches for use in abandoned object detection, and propose an abandoned object detection routine that utilises multiple modalities. To aid in the tracking and fusion of the modalities we propose a modified condensation filter that can dynamically change the particle count and features used according to the needs of the system. ----- We compare tracking and abandoned object detection performance for the proposed fusion schemes and the visual and thermal domains on their own. Testing is conducted using the OTCBVS database to evaluate object tracking, and data captured in-house to evaluate the abandoned object detection. Our results show that significant improvement can be achieved, and that a middle fusion scheme is most effective.