نوع مقاله : علمی- پژوهشی
نویسندگان
1 دانشجوی دکتری مدیریت جنگل، دانشکده منابع طبیعی، دانشگاه لرستان، خرمآباد، ایران
2 نویسنده مسئول، دانشیار، گروه مدیریت جنگل، دانشکده منابع طبیعی، دانشگاه لرستان، خرمآباد، ایران
3 دانشیار، گروه فتوگرامتری و سنجش از دور، دانشکده مهندسی نقشهبرداری، دانشگاه صنعتی خواجه نصیرالدین طوسی، تهران، ایران
4 دانشیار، گروه مدیریت جنگل، دانشکده منابع طبیعی، دانشگاه لرستان، خرمآباد، ایران
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Background and objectives: Laser scanners and digital cameras have enabled the creation of 3D models in forest environments by generating point clouds. While laser scanning technologies are often the preferred method for rendering the forest understory in three dimensions, a major limitation is their dependence on access to Global Navigation Satellite Systems (GNSS), which can complicate data acquisition in certain environments. An alternative to this limitation is the use of photogrammetry techniques. The objective of this study was to evaluate the feasibility of using high-precision terrestrial close-range photogrammetry as an alternative to traditional methods for measuring forest trees.
Methodology: Field measurements were conducted on 10 randomly selected coniferous trees located on the campus of K.N. Toosi University of Technology in Tehran, Iran, during the summer of 2022. Due to the lack of smoothness on the tree stems, image processing can lack sufficient detail. Therefore, coded and hand-made refractor targets were installed to scale the images to real-world dimensions. Control lengths were designed for orientation, and check lengths were implemented to assess accuracy. Target distances were measured using calipers and installed at various heights around the stem. Photographs were captured using a Fujifilm FinePix Real 3D W1 camera from 45 stations in stop-and-go mode, rotating 360 degrees around each tree. Stereo pair images were taken using manual focus, without a tripod. The best images were selected for each project, and matching of control and check length points was performed. One-third of the control lengths were designated as check lengths to avoid influencing the calibration. To evaluate accuracy, the following metrics were used: Root Mean Square Error (RMSE), percentage RMSE (%RMSE), Mean Absolute Error (MAE), and percentage MAE (%MAE).
Results: The accuracy of both control and check points achieved sub-millimeter precision. Total control point error was below 1 mm (0.6 mm), with RMSE = 16.79% and MAE = 10.3%. For individual trees, image reprojection errors were also analyzed in RMS pixel units. Tree number 10 showed the lowest RMS error (0.376 pixels), while tree number 5 had the highest (0.695 pixels). However, this metric alone is insufficient for full accuracy assessment, as outliers were likely retained to avoid bias in error estimation.
Conclusion: Achieving sub-millimeter accuracy is rare in forest science photogrammetry but common in industrial applications. The findings demonstrate that close-range terrestrial photogrammetry, when combined with proper network design and optimal image-capturing distance, can yield high-quality models at relatively low cost. This method shows strong potential to compete with more expensive technologies for generating dense and precise point clouds of individual trees.
کلیدواژهها [English]