Study on 3-D stereo of Marine Features on UAV High Resolution Remote Sensing Images

UAV remote sensing (UAV-RS) has been applied in many fields due to its advantages of high temporal and spatial resolution, but recent research is mainly focused on two-dimensional applications such as image registration, mosaic, classification and so on. Creatively, this study focuses on using UAV images for three-dimensional applications. First, the SIFT (Scale-invariant feature transform) algorithm is employed to automatically detect corresponding feature points between two adjacent images, and then Multiple-View Geometry and the theory of binocular stereo vision are used for matching of the image and for 3-D stereo processing of the image. Finally, the two images are synthesized for an enhanced display with three-dimensional display devices. Developed in the VC program environment, this study builds its prototype system, which achieves a clear threedimensional visualization under each of the three 3-D modes, including Anaglyphic 3D, Polarization 3D, and Active Shutter 3D. The results show that 3D stereo processing brings structured details, which provides a new way for feature interpretation and marine management.