Thesis: Ph. D. in Mechanical and Oceanographic Engineering, Joint Program in Oceanography/Applied Ocean Science and Engineering (Massachusetts Institute of Technology, Department of Mechanical Engineering; and the Woods Hole Oceanographic Institution), 2013. / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 117-135). / A fundamental problem in autonomous underwater robotics is the high latency between the capture of image data and the time at which operators are able to gain a visual understanding of the survey environment. Typical missions can generate imagery at rates hundreds of times greater than highly compressed images can be transmitted acoustically, delaying that understanding until after the vehicle has been recovered and the data analyzed. While automated classification algorithms can lessen the burden on human annotators after a mission, most are too computationally expensive or lack the robustness to run in situ on a vehicle. Fast algorithms designed for mission-time performance could lessen the latency of understanding by producing low-bandwidth semantic maps of the survey area that can then be telemetered back to operators during a mission. This thesis presents a lightweight framework for processing imagery in real time aboard a robotic vehicle. We begin with a review of pre-processing techniques for correcting illumination and attenuation artifacts in underwater images, presenting our own approach based on multi-sensor fusion and a strong physical model. Next, we construct a novel image pyramid structure that can reduce the complexity necessary to compute features across multiple scales by an order of magnitude and recommend features which are fast to compute and invariant to underwater artifacts. Finally, we implement our framework on real underwater datasets and demonstrate how it can be used to select summary images for the purpose of creating low-bandwidth semantic maps capable of being transmitted acoustically. / by Jeffrey W. Kaeli. / Ph. D. in Mechanical and Oceanographic Engineering

This thesis makes significant contributions to the object detection problem in computer vision. The object detection problem is, given a digital image of a scene, to detect the relevant object in the image. One technique for performing object detection, called ``active contours,' optimizes a constructed energy that is defined on contours (closed curves) and is tailored to image features. An optimization method can be used to perform the optimization of the energy, and thereby deform an initially placed contour to the relevant object. The typical optimization technique used in almost every active contour paper is evolving the contour by the energy's gradient descent flow, i.e., the steepest descent flow, in order to drive the initial contour to (hopefully) the minimum curve. The problem with this technique is that often times the contour becomes stuck in a sub-optimal and undesirable local minimum of the energy. This problem can be partially attributed to the fact that the gradient flows of these energies make use of only local image and contour information. By local, we mean that in order to evolve a point on the contour, only information local to that point is used. Therefore, in this thesis, we introduce a new class of flows that are global in that the evolution of a point on the contour depends on global information from the entire curve. These flows help avoid a number of problems with traditional flows including helping in avoiding undesirable local minima. We demonstrate practical applications of these flows for the object detection problem, including applications to both image segmentation and visual object tracking.

Image classification is a fundamental information extraction procedure in remote sensing that is used in land-cover and land-use mapping. Despite being considered as a replacement for manual mapping, it still requires some degree of analyst intervention. This makes the process of image classification time consuming, subjective, and error prone. For example, in unsupervised classification, pixels are automatically grouped into classes, but the user has to manually label the classes as one land-cover type or another. As a general rule, the larger the number of classes, the more difficult it is to assign meaningful class labels. A fully automated post-classification procedure for class labeling was developed in an attempt to alleviate this problem. It labels spectral classes by matching their spectral characteristics with reference spectra. A Landsat TM image of an agricultural area was used for performance assessment. The algorithm was used to label a 20- and 100-class image generated by the ISODATA classifier. The 20-class image was used to compare the technique with the traditional manual labeling of classes, and the 100-class image was used to compare it with the Spectral Angle Mapper and Maximum Likelihood classifiers. The proposed technique produced a map that had an overall accuracy of 51%, outperforming the manual labeling (40% to 45% accuracy, depending on the analyst performing the labeling) and the Spectral Angle Mapper classifier (39%), but underperformed compared to the Maximum Likelihood technique (53% to 63%). The newly developed class-labeling algorithm provided better results for alfalfa, beans, corn, grass and sugar beet, whereas canola, corn, fallow, flax, potato, and wheat were identified with similar or lower accuracy, depending on the classifier it was compared with. / vii, 93 leaves : ill., maps (some col.) ; 29 cm

[Truncated abstract] In this thesis, we introduce a mammogram analysis system developed for the automatic segmentation and analysis of mammograms. This original system has been designed to aid radiologists to detect breast cancer on mammograms. The system embodies attribute-driven segmentation in which the attributes of an image are extracted progressively in a step-by-step, hierarchical fashion. Global, low-level attributes obtained in the early stages are used to derive local, high-level attributes in later stages, leading to increasing refinement and accuracy in image segmentation and analysis. The proposed system can be characterized as: • a bootstrap engine driven by the attributes of the images; • a solid framework supporting the process of hierarchical segmentation; • a universal platform for the development and integration of segmentation and analysis techniques; and • an extensible database in which knowledge about the image is accumulated. Central to this system are three major components: 1. a series of applications for attribute acquisition; 2. a standard format for attribute normalization; and 3. a database for attribute storage and data exchange between applications. The first step of the automatic process is to segment the mammogram hierarchically into several distinctive regions that represent the anatomy of the breast. The adequacy and quality of the mammogram are then assessed using the anatomical features obtained from segmentation. Further image analysis, such as breast density classification and lesion detection, may then be carried out inside the breast region. Several domain-specific algorithms have therefore been developed for the attribute acquisition component in the system. These include: 1. automatic pectoral muscle segmentation; 2. adequacy assessment of positioning and exposure; and 3. contrast enhancement of mass lesions. An adaptive algorithm is described for automatic segmentation of the pectoral muscle on mammograms of mediolateral oblique (MLO) views

Documentation and formalization of informal settlements ("insitu" i.e. while people continue to live in the settlement) needs appropriate mapping and registration system of real property that can finally lead into integrating an informal city to the formal city. For many years extraction of geospatial data for informal settlement upgrading have been through the use of conventional mapping, which included manual plotting from aerial photographs and the use of classical surveying methods that has proved to be slow because of manual operation, very expensive, and requires well-trained personnel. The use of high-resolution satellite image like QuickBird and GIS tools has recently been gaining popularity to various aspects of urban mapping and planning, thereby opening-up new opportunities for efficient management of rapidly changing environment of informal settlements. This study was based on Manzese informal area in the city of Dar es salaam, Tanzania for which the Ministry of Lands and Human Settlement Development is committed at developing strategic information and decision making tools for upgrading informal areas using digital database, Orthophotos and Quickbird satellite image. A simple prototype approach developed in this study, that is, 'automatic detection and extraction of informal buildings and other urban features', is envisaged to simplify and speedup the process of land cover mapping that can be used by various governmental and private segments in our society. The proposed method, first tests the utility of high resolution QuickBird satellite image to classify the detailed 11 classes of informal buildings and other urban features using different image classification methods like the Box, maximum likelihood and minimum distance classifier, followed by segmentation and finally editing of feature outlines. The overall mapping accuracy achieved for detailed classification of urban land cover was 83%. The output demonstrates the potential application of the proposed approach for urban feature extraction and updating. The study constrains and recommendations for future work are also discussed. / Thesis (M.Env.Dev.)-University of KwaZulu-Natal, Pietermaritzburg, 2005.