Mean shift based clustering of Hough domain for fast line segment detection

Enhancing the city-level thermal environment through the strategic utilization of urban green spaces employing geospatial techniques

Smart urban planning needs to have a multicriteria-based approach to prevent the deteriorating local thermal climate. Maximizing the cooling potential using the available grey infrastructure would be the utmost priority of future smart cities. Remote sensing and GIS can be the appropriate tools to develop a climate-resilient urban planning framework. Studies are needed to include different features of vertical and horizontal landscaping to mitigate heat stress and enhance liveability at the city level. With this goal, the current work outlined a holistic approach to efficiently using green spaces with minimal reconstruction. The problem of regional climate threat was evaluated with urban heat island characterization. Moran's I clustering identified nearly 12% of the study area to be under considerable heat stress during summer days. Multiple techniques, such as mapping local climate zones, segment mean shift-based roof extraction, vegetation index computation, solar azimuth-based green wall site selection, etc., were applied to formulate solutions and provide an integrated method for city-level environment enhancement. A considerable area was identified as most suitable for green roof cover, and it was also computed that the transition towards green roof at only these locations may bring down the maximum heat island intensity by 0.74 °C. Additionally, solar zenith, illumination effect, and building height information were combined to create a distinct method where vertical plantation would flourish exceptionally. A rigorous assessment of more than 130 urban green spaces further quantified the relation between landscape geometry and cooling effect to provide optimum green space designs for future urban planning.

Regularized Multitask Learning for Multidimensional Log-Density Gradient Estimation

Log-density gradient estimation is a fundamental statistical problem and possesses various practical applications such as clustering and measuring nongaussianity. A naive two-step approach of first estimating the density and then taking its log gradient is unreliable because an accurate density estimate does not necessarily lead to an accurate log-density gradient estimate. To cope with this problem, a method to directly estimate the log-density gradient without density estimation has been explored and demonstrated to work much better than the two-step method. The objective of this letter is to improve the performance of this direct method in multidimensional cases. Our idea is to regard the problem of log-density gradient estimation in each dimension as a task and apply regularized multitask learning to the direct log-density gradient estimator. We experimentally demonstrate the usefulness of the proposed multitask method in log-density gradient estimation and mode-seeking clustering.

Algorithms and a Framework for Indoor Robot Mapping in a Noisy Environment Using Clustering in Spatial and Hough Domains

Map generation by a robot in a cluttered and noisy environment is an important problem in autonomous robot navigation. This paper presents algorithms and a framework to generate 2D line maps from laser range sensor data using clustering in spatial (Euclidean) and Hough domains in noisy environments. The contributions of the paper are: (1) it shows the applicability of density-based clustering methods and mathematical morphological techniques generally used in image processing for noise removal from laser range sensor data; (2) it presents a new algorithm to generate straight-line maps by applying clustering in the spatial domain; (3) it presents a new algorithm for robot mapping using clustering in a Hough domain; and (4) it presents a new framework to load, delete, install or update appropriate kernels in the robot remotely from the server. The framework provides a means to select the most appropriate kernel and fine-tune its parameters remotely from the server based on online feedback, which proves to be very efficient in dynamic environments with noisy conditions. The accuracy and performance of the techniques presented in this paper are discussed with conventional line segment-based EKF-SLAM and the results are compared.

Statistical Hough Transform

The Standard Hough Transform is a popular method in image processing and is traditionally estimated using histograms. Densities modeled with histograms in high dimensional space and/or with few observations, can be very sparse and highly demanding in memory. In this paper, we propose first to extend the formulation to continuous kernel estimates. Second, when dependencies in between variables are well taken into account, the estimated density is also robust to noise and insensitive to the choice of the origin of the spatial coordinates. Finally, our new statistical framework is unsupervised (all needed parameters are automatically estimated) and flexible (priors can easily be attached to the observations). We show experimentally that our new modeling encodes better the alignment content of images.

Curvature-based environment description for robot navigation using laser range sensors

This work proposes a new feature detection and description approach for mobile robot navigation using 2D laser range sensors. The whole process consists of two main modules: a sensor data segmentation module and a feature detection and characterization module. The segmentation module is divided in two consecutive stages: First, the segmentation stage divides the laser scan into clusters of consecutive range readings using a distance-based criterion. Then, the second stage estimates the curvature function associated to each cluster and uses it to split it into a set of straight-line and curve segments. The curvature is calculated using a triangle-area representation where, contrary to previous approaches, the triangle side lengths at each range reading are adapted to the local variations of the laser scan, removing noise without missing relevant points. This representation remains unchanged in translation or rotation, and it is also robust against noise. Thus, it is able to provide the same segmentation results although the scene will be perceived from different viewpoints. Therefore, segmentation results are used to characterize the environment using line and curve segments, real and virtual corners and edges. Real scan data collected from different environments by using different platforms are used in the experiments in order to evaluate the proposed environment description algorithm.

A BRANCH AND BOUND ALGORITHM FOR FINDING THE MODES IN KERNEL DENSITY ESTIMATES

Kernel density estimators are established tools in nonparametric statistics. Due to their flexibility and ease of use, these methods are popular in computer vision and pattern recognition for tasks such as object tracking in video or image segmentation. The most frequently used algorithm for finding the modes in such densities (the mean shift) is a gradient ascent rule, which converges to local maxima. We propose a novel, globally optimal branch and bound algorithm for finding the modes in kernel densities. We show in experiments on datasets up to dimension five that the branch and bound method is faster than local optimization and observe linear runtime scaling of our method with sample size. Quantitative experiments on simulated data show that our method gives statistically significantly more accurate solutions than the mean shift. The mode localization accuracy is about five times more precise than that of the mean shift for all tested parameters. Applications to color image segmentation on an established benchmark test set also show measurably improved results when using global optimization.