Comparison of two different photo protocols and increasing the accuracy of 3D modeling of snow shoeprints by Photogrammetry




Sheen Vento, Karla

Journal Title

Journal ISSN

Volume Title



Sometimes a shoeprint can help in narrowing down the number of suspects in a crime scene, so having an efficient recovery method for them can be helpful (Andalo et al., 2012). Photogrammetry has been proposed as a simple and reliable method for shoeprint analysis in previous studies, however, its use in certain surfaces such as snow can be challenging, and it also requires following a strict protocol for picture taking (Larsen and col., 2020). The objectives of this study are to test the equivalency of an alternative picture-taking protocol proposed by Larsen et al. with the standard protocol proposed by the developers of Digtrace, a software that allows the 3D modeling of shoeprints; and to test the effectiveness of different techniques for improving the quality of shoeprint’s photos taken in snow. In the first experiment, two shoeprints were created on sand and mud, and photographed using Larsen’s and Digtrace’s photo taking protocol. A series of 3D models were created in Digtrace, randomized, and cloud points extracted from them were compared using the CloudCompare software to assess differences in variability. In the second experiment five shoeprints were created in snow and several enhancing techniques (oblique light, red filter /black – white photo, red, and blue dyes) were used to increase the contrast of the photographs. The same comparison process from experiment was used to determine a reduction in the variability of cloud point distances with a control group. The results shown a higher accuracy from Larsen’s protocol (mean distance 0.1025 mm) than Digtrace’s protocol on mud surface, however on sand surface Digtrace’s protocol revealed less error distance (0.0968 mm) than Larsen’s protocol. The results from the second experiment shown that the use of blue and red dyes produced noticeable improvement of the reliability values. (mean error distance 0.0648 mm and 0.0734 mm). In contrast, oblique lights and red filters/black-white photos did not produce a significant improvement. This study shows that both Larsen’s and Digtrace’s protocols can be used to build reliable shoeprint 3D models and that the accuracy of 3D snow shoeprints can be improved with a simple method such as the spraying of red or blue dyes.



Shoeprint, Photogrammetry, Digitrace