The ability to detect crop rows and release sensors in large areas to ensure homogeneous coverage is crucial to monitor and increase the yield of crop rows. Aerial robotics in the agriculture field helps to reduce soil compaction. We report a release mechanics system based on image processing for crop row detection, which is essential for field navigation-based machine vision since most plants grow in a row. The release mechanics system is fully automated using embedded hardware and operated from a UAV. Once the crop row is detected, the release mechanics system releases lightweight, flexible multi-sensory devices on top of each plant to monitor the humidity and temperature conditions. The capability to monitor the local environmental conditions of plants can have a high impact on enhancing the plant’s health and in creasing the output of agriculture. The proposed algorithm steps: image acquisition, image processing, and line detection. First, we select the Region of Interest (ROI) from the frame, transform it to grayscale, remove noise, and then skeletonize and remove the background. Next, apply a Hough transform to detect crop rows and filter the lines. Finally, we use the Kalman filter to predict the crop row line in the next frame to improve the performance. This work’s main contribution is the release mechanism integrated with embedded hardware with a high-performance crop row detection algorithm for field navigation. The experimental results show the algorithm’s performance achieved a high accuracy of 90% of images with resolutions of (900x470) the speed reached 2 Frames Per Second (FPS).

No description available.

<p>In this work, we presented a novel approach to the mathematical representation of facial pose, followed by the design of a neural network (NN) capable of leveraging these representations to solve the task of facial pose estimation. Our core contribution lay in the development of advanced mathematical representations for face orientation, which include: 1) three column-vector-based representation, 2) an Anisotropic Spherical Gaussian (ASG)-based Label Distribution Learning (LDL) representation, and 3) the SO(3) Hopf coordinate-based LDL representation. These representations provided continuous and unique descriptions of the facial orientation and avoided the Gimbal lock issue of Euler angles and the antipodal issue of quaternions. Building upon these mathematical representations, we specifically designed neural network architectures to utilize these features. Key components of our NN design included 1) orthogonal loss function for column-vector-based representations which encouraged the orthogonality of predicted vectors. 2) dynamic distribution parameter learning for ASG- and SO(3)-based LDL representations which allowed the NN to adjust the contributions of adjacent labels adaptively. Our proposed mathematical representations of rotations, combined with our NN architectures, provided a powerful framework for robust and accurate facial pose estimation.</p> <p><br></p>

Stereo Image Super-Resolution (stereoSR) has attracted significant attention in recent years due to the extensive deployment of dual cameras in mobile phones, autonomous vehicles and robots. In this work, we propose a new StereoSR method, named SwinFSR, based on an extension of SwinIR, originally designed for single image restoration, and the frequency domain knowledge obtained by the Fast Fourier Convolution (FFC). Specifically, to effectively gather global information, we modify the Residual Swin Transformer blocks (RSTBs) in SwinIR by explicitly incorporating the frequency domain knowledge using the FFC and employing the resulting residual Swin Fourier Transformer blocks (RSFTBlocks) for feature extraction. Besides, for the efficient and accurate fusion of stereo views, we propose a new cross-attention module referred to as RCAM, which achieves highly competitive performance while requiring less computational cost than the state-of-the-art cross-attention modules. Extensive experimental results and ablation studies demonstrate the effectiveness and efficiency of our proposed SwinFSR. iv / Thesis / Master of Applied Science (MASc)

No description available.

This paper presents a generic pre-processor for expediting conventional template matching techniques. Instead of locating the best matched patch in the reference image to a query template via exhaustive search, the proposed algorithm rules out regions with no possible matches with minimum computational efforts. While working on simple patch features, such as mean, variance and gradient, the fast pre-screening is highly discriminative. Its computational efficiency is gained by using a novel octagonal-star-shaped template and the inclusion-exclusion principle to extract and compare patch features. Moreover, it can handle arbitrary rotation and scaling of reference images effectively, and also be robust to uniform illumination changes. GPU-aided implementation shows great efficiency of parallel computing in the algorithm design, and extensive experiments demonstrate that the proposed algorithm greatly reduces the search space while never missing the best match. / Thesis / Master of Applied Science (MASc)

A system capable of adjusting a computer vision system to unpredictable ambient lighting has been designed and attached to a silhouette robot vision system. Its principle of operation is based on the generation and analysis of the distribution of light in one T.V. frame. Designed to be used in robot vision applications, high speed processing of data is achieved in the system to generate a histogram of grey levels in one frame time. An addressable RAM technique for this purpose is explained. The system obtains two threshold values from the histogram of grey levels and places them into a threshold logic unit. A silhouette from a grey level picture is obtained as the result of the process. Adaptability of the system is performed by using different integration times in the read out of the visual transducer. The implementation of the system is based on a video rate histogram generator, a sensitivity control unit, a DMA circuit, an 86/12A microcomputer and a solid state T.V. camera. A graphics printer is used to print out results and a CRT terminal to communicate with the microcomputer. The custom hardware and software implementations for the system are depicted in detail. / Thesis / Master of Engineering (ME)

No description available.

Computer vision is concerned with the recovery of useful scene or camera information from a set of images. One classical problem is the estimation of the 3D scene structure depicted in multiple photographs. Such estimation fundamentally requires determining how the cameras are related in space. For a dynamic event recorded by multiple video cameras, finding the temporal relationship between cameras has a similar importance. Estimating such synchrony is key to a further analysis of the dynamic scene components. Existing approaches to synchronisation involve using visual cues common to both videos, and consider a discrete uniform range of synchronisation hypotheses. These prior methods exploit known constraints which hold in the presence of synchrony, from which both a temporal relationship, and an unchanging spatial relationship between the cameras can be recovered. This thesis presents methods that synchronise a pair of independently moving cameras. The spatial configuration of cameras is assumed to be known, and a cost function is developed to measure the quality of synchrony even for accuracies within a fraction of a frame. A Histogram method is developed which changes the approach from a consideration of multiple synchronisation hypotheses, to searching for seemingly synchronous frame pairs independently. Such a strategy has increased efficiency in the case of unknown frame rates. Further savings can be achieved by reducing the sampling rate of the search, by only testing for synchrony across a small subset of frames. Two robust algorithms are devised, using Bayesian inference to adaptively seek the sampling rate that minimises total execution time. These algorithms have a general underlying premise, and should be applicable to a wider class of robust estimation problems. A method is also devised to robustly synchronise two moving cameras when their spatial relationship is unknown. It is assumed that the motion of each camera has been estimated independently, so that these motion estimates are unregistered. The algorithm recovers both a synchronisation estimate, and a 3D transformation that spatially registers the two cameras. / Thesis (Ph.D.) - University of Adelaide, School of Computer Science, 2008

In this work, I examine the problem of understanding American football in video. In particular, I present several mid-level computer vision algorithms that each accomplish a different sub-task within a larger system for annotating, interpreting, and analyzing collections of American football video. The analysis of football video is useful in its own right, as teams at all levels from high school to professional football currently spend thousands of dollars and countless human work hours processing video of their own play and the play of their opponents with the aim of developing strategy and improving performance. However, because football is an extremely challenging visual domain, with difficulties ranging from the chaotic motion and identical appearance of the players to the visual clutter on the field in the form of logos and other markings, computer vision algorithms developed towards the end goal of understanding American football are broadly applicable across a variety of visual problems. I address four specific football-related problems in this thesis. First, I describe an approach for registering video with a static model (i.e. the football field in the American football domain) using a novel concept of locally distinctive invariant image feature matches. I also introduce a novel empirical registration transform stability test, which we use to initialize our registration procedure. Second, I outline a novel method for constructing mosaics from collections of video. This method takes a greedy utility maximization approach to build mosaics that achieve user-definable mosaic quality objectives. While broadly applicable, our mosaicing approach accomplishes several tasks specifically relevant to the analysis of football video, including automatically constructing reference image sets for our video registration procedure and for computing background models for initial formation recognition and player tracking algorithms. Third, I present an approach for recognizing initial player formations. This approach, called the Mixture-of-Parts Pictorial Structure (MoPPS) model, extends classical pictorial structures to recognize multi-part objects whose parts can vary in both type and location and for which an object part's location can depend on its type. While this model is effective in the American football domain, it is also broadly applicable. Finally, I address the problem of tracking football players through video using a novel particle filtering formulation and an associated discriminative training procedure that directly maximizes filter performance based on observed errors during tracking. This particle filtering framework and training procedure are also broadly applicable. For each of these algorithms, I also present a series of detailed experiments demonstrating the method's effectiveness in the American football domain. As a further contribution, I have made the data sets from most of these experiments publicly available. / Graduation date: 2013