Browsing by Subject "Feature extraction"
Now showing 1 - 3 of 3
- Results Per Page
- Sort Options
Item Catchment topography : improving hydrologic predictions with lidar analysis(2015-08) Sangireddy, Harish; Passalacqua, Paola; Maidment, David R; Hodges, Ben R; Johnson, Joel P; Liljestrand, HowardChannels, floodplains, hillslopes, and ridges are characteristic topographic features of landscapes around us. These topographic features occur at a variety of spatial scales. Climate, vegetation, soil type, and terrain characteristics control the shape of a catchment and of the channel network. Increasingly extreme and unpredictable weather patterns demand for better prediction of catchment hydrologic responses. The key to predict catchment response lies in understanding the topographic patterns and how they are influenced by the underlying processes, climate, and anthropogenic modifications. With the availability of high resolution topographic data, the characterization of topographic features at the scales relevant to hydrology and geomorphic processes is now possible. Light Detection and Ranging (lidar) digital terrain models (DTMs) (meter and sub-meter resolution) allow us to accurately quantify patterns of landscape dissection (e.g., drainage density), channel head locations, surface runoff patterns and hillslope length scales. Coarse resolution datasets (30- 100m), such as Shuttle Radar Topographic Mission (SRTM), fail to capture local variability at relevant process scales and only resolve large scale topographic patterns. As we continue to collect high resolution data there is a growing need to develop new methods and algorithms to objectively extract topographic features, such as channels, and identify metrics able to characterize topography over large areas. The goals of the research presented here are to (i) identify the signature of climate, vegetation, topography and lithology on channel patterns, (ii) define new metrics to quantify catchment topography across a range of scales, (iii) improve existing feature extraction techniques for channel networks to upscale them to handle larger catchments, and, (iv) develop feature extraction tools for urban and highly engineered setting. The research will deepen our understanding about the effects of climate on channel patterns across a range of scales; the identification of new metrics will help characterize landscapes in an objective manner; improvement of existing feature extraction techniques to handle large catchments will help in designing best management practices for watersheds through distributed mapping of topographic attributes such as slope, curvature, and accumulation area; feature extraction in urban and engineered settings will improve the analysis of watersheds modified by humans.Item Frugal Forests : learning a dynamic and cost sensitive feature extraction policy for anytime activity classification(2017-05) Kelle, Joshua Allen; Grauman, Kristen Lorraine, 1979-Many approaches to activity classification use supervised learning and so rely on extracting some form of features from the video. This feature extraction process can be computationally expensive. To reduce the cost of feature extraction while maintaining acceptable accuracy, we provide an anytime framework in which features are extracted one by one according to some policy. We propose our novel Frugal Forest feature extraction policy which learns a dynamic and cost sensitive ordering of the features. Cost sensitivity allows the policy to balance features’ predictive power with their extraction cost. The tree-like structure of the forest allows the policy to adjust on the fly in response to previously extracted feature values. We show through several experiments that the Frugal Forest policy exceeds or matches the classification accuracy per unit time of several baselines, including the current state of the art, on two challenging datasets and a variety of feature spaces.Item Localization using natural landmarks off-field for robot soccer(2013-12) He, Yuchen; Stone, Peter, 1971-Localization is an important problem that must be resolved in order for a robot to make an estimation of its location based on observation and odometry updates. Relying on artificial landmarks such as the lines, circles, and goalposts in the robot soccer domain, current robot localization requires prior knowledge and suffers from uncertainty problems due to partial observation, and thus is less generalizable compared to human beings, who refer to their surroundings for complimentary information. To improve the certainty of the localization model, we propose a framework that recognizes orientation by actively using natural landmarks from the off-field surroundings, extracting these visual features from raw images. Our approach involves identifying visual features and natural landmarks, training with localization information to understand the surroundings, and prediction based on matching of features. This approach can increase the precision of robot orientation and improve localization accuracy by eliminating uncertain hypotheses, and in addition, it is also a general approach that can be extended and applied to other localization problems as well.