The use of natural markers can improve the consistent and forgery-proof traceability of assets in sustainable supply chains. When using fingerprint technology, the inherent surface structure of an asset is captured by an imaging system and then compressed into a digital asset fingerprint which is stored in a database. To identify an asset, a newly generated fingerprint is compared against those in the database. This approach is particularly relevant for enhancing consistent asset traceability to meet regulatory requirements and support initiatives such as the digital product passport. To consistently locate and align an asset’s fingerprint region of interest, this work investigates the use of bounding symbols. Five bounding symbol shapes are empirically evaluated for their recognition performance. Additionally, four open source algorithms (ORB, BRISK, SIFT and pHash DCT) to create asset fingerprints are compared and evaluated in terms of identification confidence, processing speed, and memory requirements to assess their feasibility for tracing assets throughout life cycles. The experimental setup examines aluminum raw castings in the form of medallions. Results reveal differences in the identification confidence and resource usage among the tested algorithms. Notably, pHash DCT is more than three orders of magnitude faster than the algorithms used for feature matching, requires the least storage space, and still provides sufficient identification confidence.