Automatic building extraction remains an open research topic in digital photogrammetry and remote sensing. While many algorithms have been proposed for building extraction, none of them solve the problem completely. This is even a greater challenge in urban areas, due to high-object density and scene complexity. Standard approaches do not achieve satisfactory performance, especially with high-resolution satellite images. This paper presents a novel framework for reliable and accurate building extraction from high-resolution panchromatic images. Proposed framework exploits the domain knowledge (spatial and spectral properties) about the nature of objects in the scene, their optical interactions and their impact on the resulting image. The steps in the approach consist of 1) directional morphological enhancement; 2) multiseed-based clustering technique using internal gray variance (IGV); 3) shadow detection; 4) false alarm reduction using positional information of both building edge and shadow; and 5) adaptive threshold based segmentation technique. We have evaluated the algorithm using a variety of images from IKONOS and QuickBird satellites. The results demonstrate that the proposed algorithm is both accurate and efficient.
Abstract-Extraction of features from images has been a goal of researchers since the early days of remote sensing. While significant progress has been made in several applications, much remains to be done in the area of accurate identification of high-level features such as buildings and roads. This paper presents an approach for detecting bridges over water bodies from multispectral imagery. The multispectral image is first classified into eight land-cover types using a majority-must-be-granted logic based on the multiseed supervised classification technique. The classified image is then categorized into a trilevel image: water, concrete, and background. Bridges are then recognized in this trilevel image by using a knowledge-based approach that exploits the spatial arrangement of bridges and their surroundings using a five-step approach. A river extraction module identifies the rivers using a recursive scanning technique and geometric constraints. Using a neighborhood operator and the knowledge of the spatial dimensions of a typical bridge, we identify the possible bridge pixels. These potential bridge pixels are then grouped into possible bridge segments based on their connectivity and geometric properties.
Image segmentation, the division of a multi-dimensional image into groups of associated pixels, is an essential step for many advanced imaging applications. Image segmentation can be performed by recursively splitting the whole image or by merging together a large number of minute regions until a specified condition is satisfied. The split-and-merge procedure of image segmentation takes an intermediate level in an image description as the starting cutest, and thereby achieves a compromise between merging small primitive regions and recursively splitting the whole images to reach the desired final cutest. The proposed segmentation approach is a split-andmerge technique. The conventional split-and-merge algorithm is lacking in adaptability to the image semantics because of its stiff quadtree-based structure. In this paper, an automatic thresholding technique based on bimodality detection approach with non-homogeneity criterion is employed in the splitting phase of the split-and-merge segmentation scheme to directly reflect the image semantics to the image segmentation results. Since the proposed splitting technique depends upon homogeneity factor, some of the split regions may or may not split properly. There should be rechecking through merging technique between the two adjacent regions to overcome the drawback of the splitting technique. A sequential-arrange-based or a minimal spanning-tree based approach, that depends on data dimensionality of the weighted centroids of all split regions for finding the pair wise adjacent regions, is introduced. Finally, to overcome the problems caused by the splitting technique, a novel merging technique based on the density ratio of the adjacent pair regions is proposed. The algorithm has been tested on several synthetic as well as real life data and the results show the efficiency of the segmentation technique.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.