Dr. Zhouxuan Xiao

Dr. Zhouxuan Xiao

RS / GIS Engineer

Oral Presentations

Three-dimensional Morphological Analysis of Martian Rocks Using Zhurong Rover NaTeCam Images

March 14, 2025

Oral Poster, EGU General Assembly 2025, Vienna, Austria

This research delves into the three-dimensional (3D) morphological characteristics of Martian rocks, utilizing high-resolution images captured by the NaTeCam of China's Zhurong rover and a novel 3D rock extraction approach. The morphological analysis encompasses a detailed examination of 6,185 cm-scale rocks, highlighting the prevalence of small rocks (length < 10 cm, width < 8 cm, height < 2 cm) that constitute approximately 75% of the total rock population in the study areas. This finding underscores the significance of small-scale features in the Martian landscape and suggests that these rocks may play a substantial role in the planet's surface dynamics.

A Workflow of 3D Rock Extraction and Shape Analysis Using Mars Rover Imagery

June 18, 2021

Oral Poster, National Planetary Science Conference, Suzhou, Jiangsu, China

We present a routine and effective workflow of 3D rock extraction and shape analysis on the Martian surface based on stereo images. First, 3D point cloud data are derived from Navcam or Pancam stereoscopic images taken by the Mars rovers through photogrammetry processing, which guarantees the correctness of the point cloud scale based on the base line between the stereo cameras. Then, rock candidates are separated from the background point cloud data using Cloth Simulation Filtering (CSF) algorithm. A least squares plane fitting and distance computing from the top point to the plane are combined in order to extract the height of the rocks with high precision. Oriented Bounding Box (OBB) algorithm is used to estimate the length and width properties of the rocks. For each rock candidate, a Triangulated Irregular Network (TIN) model is generated to calculate the volume and projected area of the rock. Finally, the shape properties of the rocks (length, width, height, volume and projected area) are collected. This systematic procedure can lay a solid foundation for the 3D Rock Extraction and Shape Analysis on the Martian surface, and also provide basic information for more in-depth geological studies.

Outcrop-scale Yardang Geometric Analysis using SfM-MVS Point Clouds in Hami Area, NW China

December 12, 2020

Oral Poster, The Third Symposium on Space Earth Science, Zhuhai, Guangdong, China

Yardangs are wind-eroded ridges usually observed in arid regions on Earth and other planets. Previous geomorphology studies of terrestrial Yardangs depended on satellite data and limited sparse fieldwork. Here we design an integrated approach based on Structure-from-Motion with multi-view stereo (SfM-MVS), using a set of photos to generate centimeter-resolution ortho-mosaics and DEMs for the exploration of whaleback Yardangs in Hami area, NW China.

Integrated Approach for Calculating the Mineral Assemblages and Oxide Percentage of Crystalline Rock based on Object-Based Image Analysis and Electron Probe Microscopic Analyzer

July 07, 2019

Oral talk, IEEE International Conference on Image, Vision and Computing (ICIVC), Xiamen, Fujian, China

Identification of minerals in thin sections using intelligent methods is a very complex and challenging task, which mineralogists and computer scientists are faced with. In order to effectively calculate the the mineral assemblages and oxide percentage of crystalline rock, we proposed an integrated approach based on object-based image analysis (OBIA) and electron probe microscopic analyzer (EPMA). This approach contains the following steps: 1) acquisition of mineral chemistry data and back-scattered electron (BSE) images; 2) image segmentation and classification; 3) calculating the oxide percentage based on the proportional relationship. It is worth noting that the optimal scale parameter for image segmentation is essential for subsequent image classification, which also defines the proportional relationship for calculating the oxide percentage. The experimental results show that the novel method has characteristics of simple extraction approach, little extraction parameter, easy implementation and accurate recognition ratio.