AN IMPROVED PHOTOCLINOMETRY TECHNIQUE FOR SURFACE RELIEF RETRIEVAL FROM IMAGES: ERROR LEVELS FOR HEIGHT AND SLOPE ESTIMATES

Abstract

Subject and Purpose. Computer simulation methods are used for investigating the errors that arise in the course of retrieval, by means of an improved photoclinometry technique, of planetary surface reliefs from sets of their photo images. The work has been aimed at evaluating the level of errors in numerically calculated heights and slopes of the reliefs, as retrieved from images with a variety of signal-to-noise ratios, also including estimates for possibly minimal errors. Methods and Methodology. The improved photoclinometry approach permits calculating the most probable relief realizations for parts of a planetary surface, proceeding from sets of their photographic images. Two optional ways for implementing the method are analyzed, namely application of an optimized Fourier transform-based filtering, or solution of Poisson’s equation within the finite- difference technique. Results. Computer experiments have demonstrated that the reliefs retrievable from photo images with the use of the improved photoclinometry methods are always qualitatively similar to real ones. In the case of calculations within the finite-difference method the level of errors in height determination made 0.21s0 to 0.27s0, where s0 stands for the root-mean-square deviation in the height of the relief being modeled. In the case of application of the Fourier analysis-based method the level of errors in the calculated heights varied between 0.86s0 and 0.33s0, while the signal-to-noise ratio for the initial images changed from 1.0 to 100. Within this version of the method the theoretical prediction for the lowest error in the calculated height varied from 0.83s0 to 0.13s0. The relief belonging to the middle portion of the area under study is always retrievable to a better accuracy, as compared with the sites adjacent to the image borders, no matter which of the two available techniques has been applied. Conclusions. The improved photoclinometry method allows retrieving surface reliefs from sets of their images, with error levels for estimates of height equaling either 0.21s0 to 0.27s0 (in the case of application of the finite difference computational technique), or 0.33s0 (if the Fourier analysis has been applied, with the signal-to-noise ratio SNR 50). It is recommended that relief retrieval were performed over sites of a larger surface area than might be strictly necessary for the purpose, since the error value estimated for the middle part of the site always turns out to be several times smaller than the error calculated over the entire area under study.