Electronic publication; full text available in PDF format; abstract in HTML format. Electronic reproduction.[Australia] :Australian Digital Theses Program,2001.

Given an unstructured collection of captioned images of cluttered scenes featuring a variety of objects, our goal is to simultaneously learn the names and appearances of the objects. Only a small fraction of local features within any given image are associated with a particular caption word, and captions may contain irrelevant words not associated with any image object. We propose a novel algorithm that uses the repetition of feature neighborhoods across training images and a measure of correspondence with caption words to learn meaningful feature configurations (representing named objects). We also introduce a graph-based appearance model that captures some of the structure of an object by encoding the spatial relationships among the local visual features. In an iterative procedure we use language (the words) to drive a perceptual grouping process that assembles an appearance model for a named object. We also exploit co-occurrences among appearance models to learn hierarchical appearance models. Results of applying our method to three data sets in a variety of conditions demonstrate that from complex, cluttered, real-world scenes with noisy captions, we can learn both the names and appearances of objects, resulting in a set of models invariant to translation, scale, orientation, occlusion, and minor changes in viewpoint or articulation. These named models, in turn, are used to automatically annotate new, uncaptioned images, thereby facilitating keyword-based image retrieval.

Given an unstructured collection of captioned images of cluttered scenes featuring a variety of objects, our goal is to simultaneously learn the names and appearances of the objects. Only a small fraction of local features within any given image are associated with a particular caption word, and captions may contain irrelevant words not associated with any image object. We propose a novel algorithm that uses the repetition of feature neighborhoods across training images and a measure of correspondence with caption words to learn meaningful feature configurations (representing named objects). We also introduce a graph-based appearance model that captures some of the structure of an object by encoding the spatial relationships among the local visual features. In an iterative procedure we use language (the words) to drive a perceptual grouping process that assembles an appearance model for a named object. We also exploit co-occurrences among appearance models to learn hierarchical appearance models. Results of applying our method to three data sets in a variety of conditions demonstrate that from complex, cluttered, real-world scenes with noisy captions, we can learn both the names and appearances of objects, resulting in a set of models invariant to translation, scale, orientation, occlusion, and minor changes in viewpoint or articulation. These named models, in turn, are used to automatically annotate new, uncaptioned images, thereby facilitating keyword-based image retrieval.

Four experiments investigated object persistence in conscious awareness as a function of the nature of the cues that permit the object to be segregated from the background, and identified. A number of factors were manipulated (cue type, [color, motion, color & motion] cue duration after object identification [1s vs 5s] and cue strength [strong vs weak]). Performance was fractionated into identification, maintenance and persistence components. The results show that (1) stronger cues yielded faster identification, and (2) persistence was independent of identification time, and (3) motion cues were associated with longer persistence than color cues. A distinction between dorsal and ventral visual pathways as used to segregate the object from the background provides one way to organize the data.

Four experiments investigated object persistence in conscious awareness as a function of the nature of the cues that permit the object to be segregated from the background, and identified. A number of factors were manipulated (cue type, [color, motion, color & motion] cue duration after object identification [1s vs 5s] and cue strength [strong vs weak]). Performance was fractionated into identification, maintenance and persistence components. The results show that (1) stronger cues yielded faster identification, and (2) persistence was independent of identification time, and (3) motion cues were associated with longer persistence than color cues. A distinction between dorsal and ventral visual pathways as used to segregate the object from the background provides one way to organize the data.

The text of Beloved will be analyzed with a Kleinian and Freudian approach in order to show how the characters see each other as good or bad objects. This essay begins with an explanation of terms and a short presentation of psychoanalysis and object relations theory. Thereafter, each main character and their relation to Beloved will be examined and discussed as well as their relation to each other.

H.264/AVC is the latest international video coding standard. It was jointly developed by the Video Coding Experts Group (VCEG) of the ITU-T and the Moving Picture Experts Group (MPEG) of ISO/IEC. The goals of this standardization effort were enhanced compression efficiency and network friendly video representation. Because H.264 includes a lot of new characteristics and offers a lot of tools for compression, it can improve the quality of the compressed image greatly. H.264/AVC provides gains in compression efficiency of up to 50\% over a wide range of bit rates and video resolutions compared to previous standards. Object-based coding is the new feature that MPEG-4 supports. The object-based coding can reduce the region of motion estimation; this will increase the speed of coding. The output frame can be combined with the object-based coding sequence and also can be synthesized with the object-based coding sequence. Taking the advantage of the H.264/AVC and Object-based coding, the coding will be faster and the sequence will be smaller. In this thesis, we adopted the H.264/AVC video coding standard to implement the object coding.

In our research, we proposed a novel invariant in 2-D image contour recognition based on Hopfield-Tank neural network. At first, we searched the feature points, the position of feature points where are included high curvature and corner on the contour. We used polygonal approximation to describe the image contour. There have two patterns we set, one is model pattern another is test pattern. The Hopfield-Tank network was employed to perform feature matching. In our results show that we can overcome the test pattern which consists of translation, rotation, scaling transformation and no matter single or occlusion pattern.

Along with emergence of embedded systems, operating systems are now widely used in various applications on environments other than the desktops and workstations, such as house electrical appliances and mobile devices. Diverse applications have different requirement on the software architectures of operating systems, They can be satisfied by adopting configurable operating systems. In this research, utilizing modulization and inter-module communication channel, we developed a software architecture configurable operating system. By configuring inside channels with interfacing and protection components, we can realize an operating system into various software architectures.

Much of the existing scholarship on the popular appeal of television emphasizes the role of content over any of the medium’s other elements. Work within the cultural studies tradition, for example, often centers the importance of specific television programs when discussing the small screen’s allure for discerning viewers. Other analyses that proclaim explicit concern for “the rhetoric of television” as a whole nevertheless tend to limit their focus to specific, recognizable elements within broadcast programming. As a result, there exists no strong theoretical perspective that helps account for an attraction to television as a medium, despite that fact that many people are familiar with instances of television reception that appear to have nothing to do with the specificity of broadcast content (i.e. collapsing in front of “the box” after a long day and watching whatever happens to be on—sometimes for hours at a time). The present study remedies this absence by proposing a rhetorical mode for the medium of television based on the psychoanalytic concept of “projective identification.” Originating in the object relations work of Melanie Klein, projective identification names a primary mechanism by which individuals manage unconscious anxieties that attend modern subjectivity. This study asserts that specific elements of the televisual apparatus in combination invite unconscious acts of projective identification from viewers. Because this invitation relieves viewers of primal anxieties and increases their attraction to the medium itself, it is appropriate to interpret projective identification in this context as an inherently rhetorical concern. This study progresses in three basic sections. The first two chapters review relevant literature in the fields of rhetoric, media, and psychoanalysis in order to propose “the rhetoric of projective identification” as a mode of address inherent to the medium of television through the second half of the 20th century. The middle three chapters then validate and extend this mode by considering three elements of the televisual medium in even greater depth: Intimacy, flow, and instances of audience activism. Finally, the conclusion of the study considers the continued utility of the proposed mode in a contemporary era marked by media convergence and technological implosion.