The harmony search algorithm was developed in 2001 as a heuristic optimisation algorithm for use in diverse optimisation problems. After its introduction it was extensively used in multiple engineering disciplines with great success. In order to demonstrate the value of harmony search in computer vision applications I developed four novel algorithms based on harmony search that efficiently solves three problems that are commonly found in computer vision, namely visual tracking, visual correspondence matching and binary image restoration. Computer vision is a large discipline that includes solving many different kinds of optimisation problems. Many of these optimisation problems are discontinuous with derivative information difficult or impossible to come by. The most common solution is to use population based statistical optimisation algorithms like the particle filter, genetic algorithms, PSO, etc. but harmony search has never been investigated as a possible alternative. This is surprising since harmony search has been shown to be superior to these methods in several other engineering disciplines. I therefore aim to show that harmony search deserves to be included in the computer vision researcher's toolbox of optimisation algorithms through the introduction of four novel algorithms based on harmony search that solve three diverse problems in computer vision. First the harmony filter (HF) is introduced as a visual tracking algorithm that is shown to be superior to the particle filter and the unscented Kalman filter (UKF) in both speed and accuracy for robust tracking in challenging situations. The directed correspondence search (DCS) algorithm is then introduced as a solution to the visual correspondence problem. Finally, two algorithms, counterpoint harmony search (CHS) and largest error first harmony search (LEFHS), are introduced for the blind deconvolution of binary images. Comparative results from these algorithms are very promising. The harmony filter was compared with the particle filter and the UKF both of which have been extensively used in visual tracking. In challenging situations consisting of rapid and erratic target movement, extended periods of total and partial occlusion and changing light conditions, the HF proved to be more accurate and faster on average than both the particle filter and the UKF. Under various conditions I show that the HF is at least 2 times faster than a UKF implementation and 4 times faster than a particle filter implementation (using 300 particles). While there are fewer algorithms specialising in the blind deconvolution of binary images, CHS and LEFHS were compared with a current state-of-the-art method and proved to be more robust to noise and more accurate. LEFHS is the only algorithm currently available that can recover a 24 x 12 binary image using blind deconvolution to 100% accuracy without putting constraints on the point spread function (blurring kernel). During the development of these algorithms several valuable insights into the inner workings of harmony search were discovered. In each application harmony search had to be adapted in a different way and with each new adaptation a deeper understanding of the advantages of harmony search is revealed. Knowing which components may be modified without degrading performance is key to adapting harmony search for use in diverse problems and allows one to use harmony search in situations it was not originally designed for without losing its superior performance. These insights and the adaptation strategies that they lead to are the main contribution of this thesis.

This research aims to improve the performance of the service of a Chemotherapy Treatment Unit by reducing the waiting time of patients within the unit. In order to fulfill the objective, initially, the chemotherapy treatment unit is deduced as an identical parallel machines scheduling problem with unequal release time and single resource. A mathematical model is developed to generate the optimum schedule. Afterwards, a Tabu search (TS) algorithm is developed. The performance of the TS algorithm is evaluated by comparing results with the mathematical model and the best results of benchmark problems reported in the literature. Later on, an additional resource is considered which converted the problem into a dual resources scheduling problem. Three approaches are proposed to solve this problem; namely, heuristics, a Tabu search algorithm with heuristic (TSHu), and Tabu search algorithm for dual resources (TSD).

The complex visual search involved in baggage screening requires operators to determine quickly whether a bag contains threatening objects that are embedded in a high degree of visual clutter. Methods for calculating visual clutter have been developed, and research has demonstrated the negative impact of clutter on search performance. The current study examined whether leveraging visual clutter information on the display during search could improve baggage screening performance above and beyond the conventional screening process. Ninety undergraduates searched x-ray images of bags for weapon items in a low fidelity baggage screening simulation; two clutter-based preview conditions displayed a limited portion of the bag to the participant before the entire bag was displayed. Eye movement data confirmed that the preview process guided the participant's attention to the corresponding previewed region. However, analysis of the baggage screening performance data showed there were no significant benefits associated with either clutter-based preview conditions compared with a control condition in which the entire bag was displayed for the duration of the trial. Thus, the results suggest that using clutter-based preview to guide visual attention does not substantially improve weapon detection performance. Despite this null effect, the current study provides additional evidence regarding the impact of visual clutter on complex search performance by demonstrating significant reductions in weapon detection accuracy and search efficiency due to increasing levels of visual clutter. Further research should explore methods for improving complex visual search by considering the negative impacts of visual clutter and ensuring that both attention guidance and object recognition processes are facilitated during search.

Search engines of today are focusing on efficiently and accurately generating search results.Yet, there is much to be explored in the way people interact with the applications and relate to the content. Individuals are commonly unique, with complex preferences, motives and expectations. Not only is it important to be sensitive to these differences, but to accommodate the extremes. Enhancing a search engine does not only rely on technological development, but to explore potential user experiences in broader perspectives - which not only gratifies the needs for information, but supports a diversity of journeys. The aim of the project is to develop an alternate search engine with different functionality based on new values that reflects contemporary needs. The result, Exposeek, is an experiential prototype supporting exploratory browsing based on principles of distributed infrastructure, transparent computation and serendipitous information. Suggestive queries, legible algorithms and augmented results provide additional insights and present an alternative way to seek and peruse the Web. / Search Engines, Interaction Design

Nowadays, TV advertising is an important part of our daily life. However, it is usually hard for organizations that produce and pay for the advertisements to confirm whether their commercials are broadcasted as required in time and frequency. Consequently, a multimedia file search problem arises and it has drawn more and more attention in the past decade. In this thesis, we propose an automatic commercial search scheme using audio fingerprinting and implement it in a PC-based application. Our commercial search algorithm is composed of two parts: one for audio feature extraction and another for database search. For the first part, although the video stream of TV broadcast contains a great deal of intuitive information, we decide to ignore it because it takes much more storage and computations to process. For the audio stream, we have to extract proper audio features which can represent its characteristics and store them in a database for identification. We choose the Normalized Spectral Subband Centroids (NSSCs) as our audio fingerprints and preprocess the known commercials to build the database. For the second part, we apply a three-step process to search for any matches as the user requests, which comprises candidate search, decision-making and time verification. This process is performed for every N1 (N1=15 in our application) frames if the search result is negative. Once a match is confirmed, we skip the frames left in the commercial and use the frame after it to start a new process. Our experiment results are satisfactory based on the commercial and TV program data in our database. Moreover, it shows that our PC-based application is robust against degradation during real broadcast and recording.

Thesis (D. Min.)--Gordon-Conwell Theological Seminary, 2006. / Includes bibliographical references (leaves 213-222).

Compilation encompasses many steps. Parsing turns the input program into a more manageable syntax tree. Verification ensures that the program makes some semblance of sense. Finally, code generation transforms the internal abstract program representation into an executable program. Compilers strive to produce the best possible programs. Optimizations are applied at nearly every level of compilation. Instruction Scheduling is one of the last compilation tasks. It is part of code generation. Instruction Scheduling replaces the internal graph representation of the program with an instruction sequence. The scheduler should produce some sequence that the hardware can execute quickly. Considering that Instruction Scheduling is an NP-Complete optimization problem, it is interesting that schedules are usually generated by a greedy, heuristic algorithm called List Scheduling. Given search-based algorithms' successes in other NP-Complete optimization domains, we ask whether search-based algorithms can be applied to Instruction Scheduling to generate superior schedules without unacceptably increasing compilation time. To answer this question, we formulate a problem description that captures practical scheduling constraints. We show that this problem is NP-Complete given modest requirements on the actual hardware. We adapt three different search algorithms to Instruction Scheduling in order to show that search is an effective Instruction Scheduling technique. The schedules generated by our algorithms are generally shorter than those generated by List Scheduling. Search-based scheduling does take more time, but the increases are acceptable for some compilation domains.

No description available.

This work discusses the algorithms and implementation of a multi-robot system for locating radioactive sources. The estimation algorithm presented in this work is able to fuse measurements collected by γ-ray spectrometers carried by an unmanned aerial and unmanned ground vehicle into a single consistent estimate of the probability distribution over the position of a point source in an environment. By constructing a set of hypotheses on the position of the point source, this method converts a non-linear problem into many independent linear ones. Since the underlying model is probabilistic, candidate paths may be evaluated by their expected reduction in uncertainty, allowing the algorithm to select good paths for vehicles to take. An initial hardware test conducted at Savannah River National Laboratory served as a proof of concept and demonstrated that the algorithm successfully locates a radioactive source in the environment, and moves the vehicle to that location. This approach also demonstrated the capability to utilize radiation data collected from an unmanned aerial vehicle to aid the ground vehicle’s exploration. Subsequent numerical experiments characterized the performance of several reward functions and different exploration algorithms in scenarios covering a range of source strengths and region sizes. These experiments demonstrated the improved performance of planning-based algorithms over the myopic method initially tested in the hardware experiments. / Doctor of Philosophy / This work discusses the use of unmanned aerial and ground vehicles to autonomously locate radioactive materials. Using radiation detectors onboard each vehicle, they are commanded to search the environment using a method that incorporates measurements as they are collected. A mathematical model allows measurements taken from different vehicles in different positions to be combined together. This approach decreases the time required to locate sources by using previously collected measurements to improve the quality of later measurements. This approach also provides a best estimate of the location of a source as data is collected. This algorithm was tested in an experiment conducted at Savannah River National Laboratory. Further numerical experiments were conducted testing different reward functions and exploration algorithms.

No description available.