Translator Disclaimer
8 November 2014 Surface shape estimation of textureless area using shape from shading for Landsat imagery
Author Affiliations +
In order to obtain the DEM terrain elevation is usually calculated by photogrammetry method, which is based on establishing the correspondence between stereo images. However it is difficult to find the salient correspondence features in textureless or difficult texture regions such us mountain, snow or desert area. To solve the problem we use the shape from shading (SFS) to extract the surface elevation of the textureless area from one image. SFS is a very ill-posed and underconstrainted problem. The primary overwhelming assumption SFS is he Lambertian-based image irradiance equation with constant albedo for each ground point. Due to the physics of imaging process, satellite images do not exhibit constant albedo properties and the distribution of the surface albedo is not uniform. To address the problems above, we develop a shape recovery algorithm for the areas of weak texture or difficult texture. Given coarse DEM in weak texture area and optical image of higher spatial resolution we estimated the albedo for each ground points using relief image calculated from coarse DEM. With the known sun zenith and azimuth angles the image irradiance map is formulated as a general static Hamilton-Jacobi equation and is solved using a fast sweeping numerical method with several ground truth elevation points as boundary conditions. Landsat 7 ETM+ band 4 image and SRTM 90m global DEM data as reference surface elevation map are used to evaluate the algorithm proposed. Experiments indicate the effectiveness of the proposed method for surface elevation estimation for textureless area.
© (2014) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Yi Zhang and Jianwei Peng "Surface shape estimation of textureless area using shape from shading for Landsat imagery", Proc. SPIE 9259, Remote Sensing of the Atmosphere, Clouds, and Precipitation V, 92591S (8 November 2014);

Back to Top