A generic implementation framework for stereo matching algorithms

Traditional area-based matching techniques make use of similarity metrics such as the Sum of Absolute Differences(SAD), Sum of Squared Differences (SSD) and Normalised Cross Correlation (NCC). Non-parametric matching algorithms such as the rank and census rely on the relative ordering of pixel values rather than the pixels themselves as a similarity measure. Both traditional area-based and non-parametric stereo matching techniques have an algorithmic structure which is amenable to fast hardware realisation. This investigation undertakes a performance assessment of these two families of algorithms for robustness to radiometric distortion and random noise. A generic implementation framework is presented for the stereo matching problem and the relative hardware requirements for the various metrics investigated.

Suitability of non-parametric stereo matching algorithms for mining automation

The mining environment, being complex, irregular and time varying, presents a challenging prospect for stereo vision. For this application, speed, reliability, and the ability to produce a dense depth map are of foremost importance. This paper assesses the suitability of a number of matching techniques for use in a stereo vision sensor for close range scenes consisting primarily of rocks. These include traditional area-based matching metrics, and non-parametric transforms, in particular, the rank and census transforms. Experimental results show that the rank and census transforms exhibit a number of clear advantages over area-based matching metrics, including their low computational complexity, and robustness to certain types of distortion.

Fast and robust stereo matching algorithms for mining automation

The mining environment, being complex, irregular and time varying, presents a challenging prospect for stereo vision. For this application, speed, reliability, and the ability to produce a dense depth map are of foremost importance. This paper evaluates a number of matching techniques for possible use in a stereo vision sensor for mining automation applications. Area-based techniques have been investigated because they have the potential to yield dense maps, are amenable to fast hardware implementation, and are suited to textured scenes. In addition, two non-parametric transforms, namely, the rank and census, have been investigated. Matching algorithms using these transforms were found to have a number of clear advantages, including reliability in the presence of radiometric distortion, low computational complexity, and amenability to hardware implementation.

Quantitative evaluation of matching methods and validity measures for stereo vision

The authors present a qualitative and quantitative comparison of various similarity measures that form the kernel of common area-based stereo-matching systems. The authors compare classical difference and correlation measures as well as nonparametric measures based on the rank and census transforms for a number of outdoor images. For robotic applications, important considerations include robustness to image defects such as intensity variation and noise, the number of false matches, and computational complexity. In the absence of ground truth data, the authors compare the matching techniques based on the percentage of matches that pass the left-right consistency test. The authors also evaluate the discriminatory power of several match validity measures that are reported in the literature for eliminating false matches and for estimating match confidence. For guidance applications, it is essential to have and estimate of confidence in the three-dimensional points generated by stereo vision. Finally, a new validity measure, the rank constraint, is introduced that is capable of resolving ambiguous matches for rank transform-based matching.

Binocular rivalry and visuospatial ability in individuals with schizophrenia

Visual abnormalities, both at the sensory input and the higher interpretive levels, have been associated with many of the symptoms of schizophrenia. Individuals with schizophrenia typically experience distortions of sensory perception, resulting in perceptual hallucinations and delusions that are related to the observed visual deficits. Disorganised speech, thinking and behaviour are commonly experienced by sufferers of the disorder, and have also been attributed to perceptual disturbances associated with anomalies in visual processing. Compounding these issues are marked deficits in cognitive functioning that are observed in approximately 80% of those with schizophrenia. Cognitive impairments associated with schizophrenia include: difficulty with concentration and memory (i.e. working, visual and verbal), an impaired ability to process complex information, response inhibition and deficits in speed of processing, visual and verbal learning. Deficits in sustained attention or vigilance, poor executive functioning such as poor reasoning, problem solving, and social cognition, are all influenced by impaired visual processing. These symptoms impact on the internal perceptual world of those with schizophrenia, and hamper their ability to navigate their external environment. Visual processing abnormalities in schizophrenia are likely to worsen personal, social and occupational functioning. Binocular rivalry provides a unique opportunity to investigate the processes involved in visual awareness and visual perception. Binocular rivalry is the alternation of perceptual images that occurs when conflicting visual stimuli are presented to each eye in the same retinal location. The observer perceives the opposing images in an alternating fashion, despite the sensory input to each eye remaining constant. Binocular rivalry tasks have been developed to investigate specific parts of the visual system. The research presented in this Thesis provides an explorative investigation into binocular rivalry in schizophrenia, using the method of Pettigrew and Miller (1998) and comparing individuals with schizophrenia to healthy controls. This method allows manipulations to the spatial and temporal frequency, luminance contrast and chromaticity of the visual stimuli. Manipulations to the rival stimuli affect the rate of binocular rivalry alternations and the time spent perceiving each image (dominance duration). Binocular rivalry rate and dominance durations provide useful measures to investigate aspects of visual neural processing that lead to the perceptual disturbances and cognitive dysfunction attributed to schizophrenia. However, despite this promise the binocular rivalry phenomenon has not been extensively explored in schizophrenia to date. Following a review of the literature, the research in this Thesis examined individual variation in binocular rivalry. The initial study (Chapter 2) explored the effect of systematically altering the properties of the stimuli (i.e. spatial and temporal frequency, luminance contrast and chromaticity) on binocular rivalry rate and dominance durations in healthy individuals (n=20). The findings showed that altering the stimuli with respect to temporal frequency and luminance contrast significantly affected rate. This is significant as processing of temporal frequency and luminance contrast have consistently been demonstrated to be abnormal in schizophrenia. The current research then explored binocular rivalry in schizophrenia. The primary research question was, "Are binocular rivalry rates and dominance durations recorded in participants with schizophrenia different to those of the controls?" In this second study binocular rivalry data that were collected using low- and highstrength binocular rivalry were compared to alternations recorded during a monocular rivalry task, the Necker Cube task to replicate and advance the work of Miller et al., (2003). Participants with schizophrenia (n=20) recorded fewer alternations (i.e. slower alternation rates) than control participants (n=20) on both binocular rivalry tasks, however no difference was observed between the groups on the Necker cube task. Magnocellular and parvocellular visual pathways, thought to be abnormal in schizophrenia, were also investigated in binocular rivalry. The binocular rivalry stimuli used in this third study (Chapter 4) were altered to bias the task for one of these two pathways. Participants with schizophrenia recorded slower binocular rivalry rates than controls in both binocular rivalry tasks. Using a ‘within subject design’, binocular rivalry data were compared to data collected from a backwardmasking task widely accepted to bias both these pathways. Based on these data, a model of binocular rivalry, based on the magnocellular and parvocellular pathways that contribute to the dorsal and ventral visual streams, was developed. Binocular rivalry rates were compared with performance on the Benton’s Judgment of Line Orientation task, in individuals with schizophrenia compared to healthy controls (Chapter 5). The Benton’s Judgment of Line Orientation task is widely accepted to be processed within the right cerebral hemisphere, making it an appropriate task to investigate the role of the cerebral hemispheres in binocular rivalry, and to investigate the inter-hemispheric switching hypothesis of binocular rivalry proposed by Pettigrew and Miller (1998, 2003). The data were suggestive of intra-hemispheric rather than an inter-hemispheric visual processing in binocular rivalry. Neurotransmitter involvement in binocular rivalry, backward masking and Judgment of Line Orientation in schizophrenia were investigated using a genetic indicator of dopamine receptor distribution and functioning; the presence of the Taq1 allele of the dopamine D2 receptor (DRD2) receptor gene. This final study (Chapter 6) explored whether the presence of the Taq1 allele of the DRD2 receptor gene, and thus, by inference the distribution of dopamine receptors and dopamine function, accounted for the large individual variation in binocular rivalry. The presence of the Taq1 allele was associated with slower binocular rivalry rates or poorer performance in the backward masking and Judgment of Line Orientation tasks seen in the group with schizophrenia. This Thesis has contributed to what is known about binocular rivalry in schizophrenia. Consistently slower binocular rivalry rates were observed in participants with schizophrenia, indicating abnormally-slow visual processing in this group. These data support previous studies reporting visual processing abnormalities in schizophrenia and suggest that a slow binocular rivalry rate is not a feature specific to bipolar disorder, but may be a feature of disorders with psychotic features generally. The contributions of the magnocellular or dorsal pathways and parvocellular or ventral pathways to binocular rivalry, and therefore to perceptual awareness, were investigated. The data presented supported the view that the magnocellular system initiates perceptual awareness of an image and the parvocellular system maintains the perception of the image, making it available to higher level processing occurring within the cortical hemispheres. Abnormal magnocellular and parvocellular processing may both contribute to perceptual disturbances that ultimately contribute to the cognitive dysfunction associated with schizophrenia. An alternative model of binocular rivalry based on these observations was proposed.

Online calibration of stereo rigs for long-term autonomy

Stereo-based visual odometry algorithms are heavily dependent on an accurate calibration of the rigidly fixed stereo pair. Even small shifts in the rigid transform between the cameras can impact on feature matching and 3D scene triangulation, adversely affecting pose estimates and applications dependent on long-term autonomy. In many field-based scenarios where vibration, knocks and pressure change affect a robotic vehicle, maintaining an accurate stereo calibration cannot be guaranteed over long periods. This paper presents a novel method of recalibrating overlapping stereo camera rigs from online visual data while simultaneously providing an up-to-date and up-to-scale pose estimate. The proposed technique implements a novel form of partitioned bundle adjustment that explicitly includes the homogeneous transform between a stereo camera pair to generate an optimal calibration. Pose estimates are computed in parallel to the calibration, providing online recalibration which seamlessly integrates into a stereo visual odometry framework. We present results demonstrating accurate performance of the algorithm on both simulated scenarios and real data gathered from a wide-baseline stereo pair on a ground vehicle traversing urban roads.

High altitude stereo visual odometry

Stereo visual odometry has received little investigation in high altitude applications due to the generally poor performance of rigid stereo rigs at extremely small baseline-to-depth ratios. Without additional sensing, metric scale is considered lost and odometry is seen as effective only for monocular perspectives. This paper presents a novel modification to stereo based visual odometry that allows accurate, metric pose estimation from high altitudes, even in the presence of poor calibration and without additional sensor inputs. By relaxing the (typically fixed) stereo transform during bundle adjustment and reducing the dependence on the fixed geometry for triangulation, metrically scaled visual odometry can be obtained in situations where high altitude and structural deformation from vibration would cause traditional algorithms to fail. This is achieved through the use of a novel constrained bundle adjustment routine and accurately scaled pose initializer. We present visual odometry results demonstrating the technique on a short-baseline stereo pair inside a fixed-wing UAV flying at significant height (~30-100m).

Robust scale initialization for long-range stereo visual odometry

Achieving a robust, accurately scaled pose estimate in long-range stereo presents significant challenges. For large scene depths, triangulation from a single stereo pair is inadequate and noisy. Additionally, vibration and flexible rigs in airborne applications mean accurate calibrations are often compromised. This paper presents a technique for accurately initializing a long-range stereo VO algorithm at large scene depth, with accurate scale, without explicitly computing structure from rigidly fixed camera pairs. By performing a monocular pose estimate over a window of frames from a single camera, followed by adding the secondary camera frames in a modified bundle adjustment, an accurate, metrically scaled pose estimate can be found. To achieve this the scale of the stereo pair is included in the optimization as an additional parameter. Results are presented both on simulated and field gathered data from a fixed-wing UAV flying at significant altitude, where the epipolar geometry is inaccurate due to structural deformation and triangulation from a single pair is insufficient. Comparisons are made with more conventional VO techniques where the scale is not explicitly optimized, and demonstrated over repeated trials to indicate robustness.

An accuracy assessment of ASTER stereo images-derived digital elevation model by using rational polynomial coefficient model

The along-track stereo images of Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) sensor with 15 m resolution were used to generate Digital Elevation Model (DEM) on an area with low and near Mean Sea Level (MSL) elevation in Johor, Malaysia. The absolute DEM was generated by using the Rational Polynomial Coefficient (RPC) model which was run on ENVI 4.8 software. In order to generate the absolute DEM, 60 Ground Control Pointes (GCPs) with almost vertical accuracy less than 10 meter extracted from topographic map of the study area. The assessment was carried out on uncorrected and corrected DEM by utilizing dozens of Independent Check Points (ICPs). Consequently, the uncorrected DEM showed the RMSEz of ± 26.43 meter which was decreased to the RMSEz of ± 16.49 meter for the corrected DEM after post-processing. Overall, the corrected DEM of ASTER stereo images met the expectations.

Echolocation call intensity in the aerial hawking bat Eptesicus bottae (Vespertilionidae) studied using stereo videogrammetry

Aerial hawking bats use intense echolocation calls to search for insect prey. Their calls have evolved into the most intense airborne animal vocalisations. Yet our knowledge about call intensities in the field is restricted to a small number of species. We describe a novel stereo videogrammetry method used to study flight and echolocation behaviour, and to measure call source levels of the aerial hawking bat Eptesicus bottae (Vespertilionidae). Bats flew close to their predicted minimum power speed. Source level increased with call duration; the loudest call of E. bottae was at 133 dB peSPL. The calculated maximum detection distance for large flying objects (e.g. large prey, conspecifics) was up to 21 m. The corresponding maximum echo delay is almost exactly the duration of one wing beat in E. bottae and this also is its preferred pulse interval. These results, obtained by using videogrammetry to track bats in the field, corroborate earlier findings from other species from acoustic tracking methods.

Long-range stereo visual odometry for unmanned aerial vehicles

This thesis explored the utility of long-range stereo visual odometry for application on Unmanned Aerial Vehicles. Novel parameterisations and initialisation routines were developed for the long-range case of stereo visual odometry and new optimisation techniques were implemented to improve the robustness of visual odometry in this difficult scenario. In doing so, the applications of stereo visual odometry were expanded and shown to perform adequately in situations that were previously unworkable.

Simultaneous underwater visibility assessment, enhancement and improved stereo

Vision-based underwater navigation and obstacle avoidance demands robust computer vision algorithms, particularly for operation in turbid water with reduced visibility. This paper describes a novel method for the simultaneous underwater image quality assessment, visibility enhancement and disparity computation to increase stereo range resolution under dynamic, natural lighting and turbid conditions. The technique estimates the visibility properties from a sparse 3D map of the original degraded image using a physical underwater light attenuation model. Firstly, an iterated distance-adaptive image contrast enhancement enables a dense disparity computation and visibility estimation. Secondly, using a light attenuation model for ocean water, a color corrected stereo underwater image is obtained along with a visibility distance estimate. Experimental results in shallow, naturally lit, high-turbidity coastal environments show the proposed technique improves range estimation over the original images as well as image quality and color for habitat classification. Furthermore, the recursiveness and robustness of the technique allows implementation onboard an Autonomous Underwater Vehicle for improving navigation and obstacle avoidance performance.

Tracking people in 3D using position, size and shape

This paper presents a prototype tracking system for tracking people in enclosed indoor environments where there is a high rate of occlusions. The system uses a stereo camera for acquisition, and is capable of disambiguating occlusions using a combination of depth map analysis, a two step ellipse fitting people detection process, the use of motion models and Kalman filters and a novel fit metric, based on computationally simple object statistics. Testing shows that our fit metric outperforms commonly used position based metrics and histogram based metrics, resulting in more accurate tracking of people.

Bearing-only SLAM : a vision-based navigation system for autonomous robots

To navigate successfully in a previously unexplored environment, a mobile robot must be able to estimate the spatial relationships of the objects of interest accurately. A Simultaneous Localization and Mapping (SLAM) sys- tem employs its sensors to build incrementally a map of its surroundings and to localize itself in the map simultaneously. The aim of this research project is to develop a SLAM system suitable for self propelled household lawnmowers. The proposed bearing-only SLAM system requires only an omnidirec- tional camera and some inexpensive landmarks. The main advantage of an omnidirectional camera is the panoramic view of all the landmarks in the scene. Placing landmarks in a lawn field to define the working domain is much easier and more flexible than installing the perimeter wire required by existing autonomous lawnmowers. The common approach of existing bearing-only SLAM methods relies on a motion model for predicting the robot’s pose and a sensor model for updating the pose. In the motion model, the error on the estimates of object positions is cumulated due mainly to the wheel slippage. Quantifying accu- rately the uncertainty of object positions is a fundamental requirement. In bearing-only SLAM, the Probability Density Function (PDF) of landmark position should be uniform along the observed bearing. Existing methods that approximate the PDF with a Gaussian estimation do not satisfy this uniformity requirement. This thesis introduces both geometric and proba- bilistic methods to address the above problems. The main novel contribu- tions of this thesis are: 1. A bearing-only SLAM method not requiring odometry. The proposed method relies solely on the sensor model (landmark bearings only) without relying on the motion model (odometry). The uncertainty of the estimated landmark positions depends on the vision error only, instead of the combination of both odometry and vision errors. 2. The transformation of the spatial uncertainty of objects. This thesis introduces a novel method for translating the spatial un- certainty of objects estimated from a moving frame attached to the robot into the global frame attached to the static landmarks in the environment. 3. The characterization of an improved PDF for representing landmark position in bearing-only SLAM. The proposed PDF is expressed in polar coordinates, and the marginal probability on range is constrained to be uniform. Compared to the PDF estimated from a mixture of Gaussians, the PDF developed here has far fewer parameters and can be easily adopted in a probabilistic framework, such as a particle filtering system. The main advantages of our proposed bearing-only SLAM system are its lower production cost and flexibility of use. The proposed system can be adopted in other domestic robots as well, such as vacuum cleaners or robotic toys when terrain is essentially 2D.

Effect of amblyopia on the developmental eye movement test in children

Purpose. To investigate the functional impact of amblyopia in children, the performance of amblyopic and age-matched control children on a clinical test of eye movements was compared. The influence of visual factors on test outcome measures was explored. Methods. Eye movements were assessed with the Developmental Eye Movement (DEM) test, in a group of children with amblyopia (n = 39; age, 9.1 ± 0.9 years) of different causes (infantile esotropia, n = 7; acquired strabismus, n = 10; anisometropia, n = 8; mixed, n = 8; deprivation, n = 6) and in an age-matched control group (n = 42; age, 9.3 ± 0.4 years). LogMAR visual acuity (VA), stereoacuity, and refractive error were also recorded in both groups. Results. No significant difference was found between the amblyopic and age-matched control group for any of the outcome measures of the DEM (vertical time, horizontal time, number of errors and ratio(horizontal time/vertical time)). The DEM measures were not significantly related to VA in either eye, level of binocular function (stereoacuity), history of strabismus, or refractive error. Conclusions. The performance of amblyopic children on the DEM, a commonly used clinical measure of eye movements, has not previously been reported. Under habitual binocular viewing conditions, amblyopia has no effect on DEM outcome scores despite significant impairment of binocular vision and decreased VA in both the better and worse eye.