Measuring antler lengths using low-cost ToF cameras
Date
2024-05-13
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Antlers have been widely used in pharmaceuticals, understanding regenerative mechanism, monitoring environmental pollutants as bio-indicators, and studying mechanical properties of bony tissues. Antler measurements play a significant role in these domains, indicating the antler growth stages, which are intricately linked to the medical properties, regenerative processes, and variations in pollutant levels within the antler. Traditionally, antler measurements are performed with contact methods like the measuring tape. The complex antler geometry entails multiple measurements and selection of reference points at various locations, making the process prone to human errors. The measuring process also requires manual capture or the use of drugs to keep the animal stationary, raising ethical and security concerns.
To address these challenges and accurately monitor antler growth without causing any harm to the animals, an optical imaging method is proposed using a multi-camera system to obtain 3D antler data. The designed imaging system incorporates multiple Time-of-flight (ToF) cameras, an RGB camera, and an external trigger. The RGB camera detects animal motion, together with the external trigger, facilitates sequential data capture by the ToF cameras when animals are not moving substantially, thereby avoiding motion blur and camera interference. This mechanism ensures the collection of complete and usable antler data. In order to generate high quality antler data suitable for extracting measurements, a highly automated data processing workflow has been designed including antler quality control, segmentation, two-step registration and antler denoising. Animal motion and the low reflectivity of antlers are major factors degrading ToF camera data quality. To mitigate motion blur, animals are scanned during stationary periods and frames collected in each static period are merged as a weighted average for improved quality. The antler point cloud is automatically extracted by searching k-nearest neighbors from the environment data. The two-step registration involves registering multiple ToF cameras and aligning antler data from various static periods, which densify the antler data for manipulation. Ultimately, a polar coordinate-based denoising process filters out blunders and noise from the antler data for subsequent modelling process. An adaptive modelling approach has been developed to mathematically represent the antler data and facilitate measurement extraction. This method breaks down the 3D point cloud into slices and reconstructs the contour of each slice. The slice-based method strategically positions feature points where curvature changes occur, allowing for efficient storage and accurate reconstruction. Remarkably, this approach extends beyond antlers and can be applied to diverse shapes without prior knowledge. In summary, this research presents a comprehensive solution for monitoring antler growth, emphasizing precision, automation, and adaptability in the modeling processes and measurement extraction. This innovative approach not only advances antler related research but also lays the foundation for similar studies involving complex biological structures.
Description
Keywords
Time-of-flight camera, Point Cloud, 3D Modeling, Antler, Skeletonization
Citation
Cheng, S. (2024). Measuring antler lengths using low-cost ToF cameras (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.