Efficient and accurate set-based registration of time-separated aerial images

This paper addresses the task of time-separated aerial image registration. The ability to solve this problem accurately and reliably is important for a variety of subsequent image understanding applications. The principal challenge lies in the extent and nature of transient appearance variation that a land area can undergo, such as that caused by the change under illumination conditions, seasonal variations, or the occlusion by non-persistent objects (people, cars). Our work introduces several major novelties (i) unlike previous work on aerial image registration, we approach the problem using a set-based paradigm; (ii) we show how image space local, pair-wise constraints can be used to enforce a globally good registration using a constraints graph structure; (iii) we show how a simple holistic representation derived from raw aerial images can be used as a basic building block of the constraints graph in a manner which achieves both high registration accuracy and speed; (iv) lastly, we introduce a new and, to the best of our knowledge, the only data corpus suitable for the evaluation of set-based aerial image registration algorithms. Using this data set, we demonstrate (i) that the proposed method outperforms the state-of-the-art for pair-wise registration already, achieving greater accuracy and reliability, while at the same time reducing the computational cost of the task and (ii) that the increase in the number of available images in a set consistently reduces the average registration error, with a major difference already for a single additional image.

Adaptive robust control of 6-DOF autonomous aerial vehicles

The thesis focused on development of an auto-pilot system for UAV’s and small fixed wing aircraft for use in hazardous flight conditions, such as severe weather. This led to development of a mathematical algorithm that unbinds the flight systems from coupling effects which can adaptively changed to the environment.

Groupwise registration of aerial images

This paper addresses the task of time separated aerial image registration. The ability to solve this problem accurately and reliably is important for a variety of subsequent image understanding applications. The principal challenge lies in the extent and nature of transient appearance variation that a land area can undergo, such as that caused by the change in illumination conditions, seasonal variations, or the occlusion by non-persistent objects (people, cars). Our work introduces several novelties: (i) unlike all previous work on aerial image registration, we approach the problem using a set-based paradigm; (ii) we show how local, pairwise constraints can be used to enforce a globally good registration using a constraints graph structure; (iii) we show how a simple holistic representation derived from raw aerial images can be used as a basic building block of the constraints graph in a manner which achieves both high registration accuracy and speed. We demonstrate: (i) that the proposed method outperforms the state-of-the-art for pair-wise registration already, achieving greater accuracy and reliability, while at the same time reducing the computational cost of the task; and (ii) that the increase in the number of available images in a set consistently reduces the average registration error.

A new perspective of storm bite on sandy beaches using unmanned aerial vehicles

Aerial imagery collected before and after major storm events is ideal for the assessment of coastal landscape change driven by individual high-magnitude events. Using traditional satellite sensors and manned aerial systems can be challenging due to issues related to cloud cover, mobilization expenses and resolution. Rapid advances in unmanned aerial vehicle (UAV) technology allow for the cost-effective collection of aerial imagery and topography at centimetre resolution suitable for assessing change in coastal ecosystems. In this study we demonstrate the utility of UAV-based photogrammetry to quantify storm-driven sediment dynamics on a sandy beach on the open-coast shoreline of Victoria, Australia. UAV-based aerial photography was collected before and after a major storm event. High-resolution (< 5 cm) aerial imagery and digital surface models were acquired and change-detection techniques were applied to quantify changes in the beachface. An average beach erosion of 12.24 m3/m with a maximum of 28.05 m3/m was observed, and the volume of sand cut from the beachface and retreat of the foredune are clearly illustrated. Following the storm event, erosion was estimated at 7259.94 ± 503.69 m3 along 550 m of beach. By combining the aerial imagery and derived topographic datasets we demonstrate the advantage of UAV-based photogrammetry techniques for rapid high-resolution data collection in semi-remote locations. Its utility in setting unlimited virtual vantage points is also illustrated and the valuable perspective it provides for tracking landscape change discussed.

Conceptual design for fully autonomous aerial and ground system for precision agriculture

There are few regulatory restrictions involving the use of fully autonomous unmanned aerial systems in unpopulated, farming areas of Australia. The combination of a fully autonomous aerial and ground systems would provide efficient and cost effective retrieval of soil and vegetation data for use in precision agriculture. The aerial system will survey the site and collect spectral imagery to analyse plant density, stress and nutrition. The ground sensors will collect soil moisture content readings throughout the site. The data from both systems will be collated at a central base station. The base station will also provide housing and interface with the aerial system.

The training load of aerial skiing

This study quantified the training load experienced by elite aerial skiers. Nine elite female aerial skiers were monitored during 16 training sessions over a 13 day period. Time-motion, landing impact and heart rate (HR) data were measured from 688 jumps using integrated GPS, accelerometer and HR transmitters while rating of perceived exertion (RPE) was taken using Borg's scale. Each jump was delineated into five components from the GPS time-motion data to determine the work to rest ratios. Participants completed 16 ± 3 jumps per session with a work to rest ratio of 1.9:1 Heart rates averaged 65 ± 3.1% HRmax and peaked at 85 ± 4.4% HRmax while and an RPE score of 12 ± 1 was evoked. Landing impacts were significantly higher (p ≤ 0.001) when participants jumped off ramps with a larger take-off angle or when they completed jumps with a mid-air rotation. The training load experienced by elite aerial skiers may be causative of the high incidence of injuries reported. Significantly differing levels of impact load during the study suggest training load for these athletes can be easily modified and periodised allowing optimised performance and minimised injury.