Jyun Ping Jhan

Analysis of Oblique Aerial Images for Land Cover and Point Cloud Classification in an Urban Environment

In addition to aerial imagery, point clouds are important remote sensing data in urban environment studies. It is essential to extract semantic information from both images and point clouds for such purposes; thus, this study aims to automatically classify 3-D point clouds generated using oblique aerial imagery (OAI)/vertical aerial imagery (VAI) into various urban object classes, such as roof, facade, road, tree, and grass. A multicamera airborne imaging system that can simultaneously acquire VAI and OAI is suggested. The acquired small-format images contain only three RGB spectral bands and are used to generate photogrammetric point clouds through a multiview-stereo dense matching technique. To assign each 3-D point cloud to a corresponding urban object class, we first analyzed the original OAI through object-based image analyses. A rule-based hierarchical semantic classification scheme that utilizes spectral information and geometry- and topology-related features was developed, in which the object height and gradient features were derived from the photogrammetric point clouds to assist in the detection of elevated objects, particularly for the roof and facade. Finally, the photogrammetric point clouds were classified into the aforementioned five classes. The classification accuracy was assessed on the image space, and four experimental results showed that the overall accuracy is between 82.47% and 91.8%. In addition, visual and consistency analyses were performed to demonstrate the proposed classification schemes feasibility, transferability, and reliability, particularly for distinguishing elevated objects from OAI, which has a severe occlusion effect, image-scale variation, and ambiguous spectral characteristics.

Semiautomatic Object-Oriented Landslide Recognition Scheme From Multisensor Optical Imagery and DEM

Rainfall-induced landslides are a major threat in Taiwan, particularly during the typhoon season. A precise survey of landslides after a super event is a critical task for disaster, watershed, and forestry land management. In this paper, we utilize high spatial resolution multispectral optical imagery and a digital elevation model (DEM) with an object-oriented analysis technique to develop a scheme for the recognition of landslides using multilevel segmentation and a hierarchical semantic network. Four case studies are presented to evaluate the feasibility of the proposed scheme. Three kinds of remote sensing imagery, namely pan-sharpened FORMOSAT-2 satellite images, aerial digital images from Z/I digital mapping camera, and images acquired by a digital single lens reflex camera mounted on a fixed-wing unmanned aerial vehicle are used. An accuracy assessment is accomplished by evaluating three test sites containing hundreds of landslides associated with the Typhoon Morakot. The input data include ortho-rectified image and DEM. Four spectral and one topographic object features are derived for semiautomatic landslide recognition. The threshold values are determined semiautomatically by statistical estimation from a few training samples. The experimental results show that the proposed approach can counteract the commission/omission errors and achieve missing/branching factors at less than 0.12 with a quality percentage of 81.7%. The results demonstrate the feasibility and accuracy of the proposed landslide recognition scheme even when different optical sensors are utilized.

Development of a Large-Format UAS Imaging System with the Construction of a One Sensor Geometry from a Multicamera Array

For the purpose of large-area topographic mapping, this study proposes an imaging system based on a multicamera array unmanned aerial system (UAS) comprised of five small-format digital cameras with a total field of view of 127°. The five digital cameras are aligned in a row along the across-track direction with overlap between two neighboring cameras. The suggested system has higher data acquisition efficiency than the single-camera UAS imaging system. For topographic mapping purposes, we develop a modified projective transformation method to stitch all five raw images into one sensor geometry. In this method, the transformation coefficients are obtained by on-the-job multicamera self-calibration, including interior and relative orientations. During the stitching process, two systematic errors are detected and corrected. In the end, a large-format digital image can be produced for each trigger event independently. The photogrammetric collinearity condition is evaluated using several external accuracy assessments, such as conventional aerial triangulation, stereoplotting, and digital surface model generation procedures. From the accuracy assessment results, we conclude that the presented raw image stitching method can be used to construct a one sensor geometry from a multicamera array and is feasible for 3-D mapping applications.