Construction of pixel-level resolution DEMs from monocular images by shape and albedo from shading constrained with low-resolution DEM

An Integrated Photogrammetric and Photoclinometric Approach for Pixel-Resolution 3d Modelling of Lunar Surface

“…integrating stereo photogrammetry and photoclinometry for high-resolution lunar surface modelling have been explored and developed (Grumpe et al, 2014;Wu et al, 2017). These algorithms mainly focus on applying photoclinometry on the derived products (e.g., a DEM) from conventional approaches such as stereo photogrammetry and laser altimetry.…”

“…3D positions of the matched conjugate points are photogrammetrically determined and an improved point cloud can be generated based on the increased number of conjugate points from photoclinometric processing. Improved pixelresolution 3D surface models can hence be obtained by using the aforementioned algorithms (Grumpe et al, 2014;Wu et al, 2017) with an improved photogrammetric 3D model from the proposed approach. This paper first describes the integrated photogrammetric and photoclinometric approach, followed by experimental analysis of the approach with real data using LROC NAC images.…”

“…As a result, all the 3D points derived at this stage are photogrammetrically determined; doing so ensures a consistent accuracy and uncertainty pattern for all computed 3D points. Afterwards, the 3D model is used as constraints in photoclinometric reconstruction algorithms such as Grumpe et al (2014) and Wu et al (2017) for pixel-resolution modelling. This strategy derives additional information (i.e., conjugate points) from photoclinometric processing and extends its application spectrum.…”

“…The proposed albedo estimation in this research is an integrated extension of the aforementioned methods. As image intensity is a ratio of incoming and reflected energy ranging from 0 to 1, assuming a pixel must be lit up by illumination (i.e., I > 0), there is a physical limit for both albedo and geometric reflectance (Wu et al, 2017):…”

“…Although there are methods to solve for both p and q in one observation (Grumpe et al, 2014;Wu et al, 2017), they usually require additional information or constraints that may lose optimal image details for matching. Instead, this paper solves for the normal (i.e., negative tangent) along the illumination direction which plausible solutions can be determined (Horn, 1990).…”