These datasets were gathered during an examination of non-identifiable fingermarks from unfired ammunition in handguns.
The files posted are datasets from a project that examined the frequency of non-identifiable fingermarks occur on naturally loaded handgun ammunition and what range of associative values can be expected. The datasets include images of 415 Non-Identifiable Fingermarks. The Non-Identifiable Fingermark Dataset has original images in .tif format, whereas the Annotated Non-Identifiable Fingermark Dataset has processed images with annotation of fingermark minutiae. The accompanying spreadsheets contain summary information. The spreadsheet Marks by Round provides the number of identifiable and non-identifiable fingermarks found, the number of annotated minutiae for the marks, the handgun type (semi-automatic vs. revolver), and the caliber class. The spreadsheet Marks by Handgun Load summarizes the results from each of the handguns sampled. The four caliber classes used were 22; 32; 38; and .45. Ammunition was collected directly from firearms loaded previously by their owners, without any specific guidance or requirements for preparation of ammunition, washing of hands, or loading technique. 1263 rounds of ammunition were collected from 164 handguns. Fingermarks were visualized on ammunition using cyanoacrylate fuming (CA, superglue), followed by fluorescent dye staining (using the dye BY-40). Photography of the cylindrical ammunition surfaces required multiple exposures around the circumference (up to 15 for the complete cartridge surface). These individual images were cropped and stitched together (analogous to panoramic photography). Images of candidate NIFMs were examined by a highly experienced certified latent print examiner and accepted where 4 or more minutiae were present and in a clear relationship to one another in the context of the ridge flow. Minutiae were documented by annotation of the images and graded as either Of Value for Identification (meeting traditional requirements for a fingerprint identification) or No Value (a non-identifiable fingermark, NIFM). No images of fingermarks graded as identifiable are in the dataset.