Selecting Valid Correlation Areas for Automated Bullet Identification System Based on Striation Detection

Some automated bullet identification systems calculate a correlation score between two land impressions to measure their similarity. When extracting a compressed profile from the land impression of a fired bullet, inclusion of areas that do not contain valid individual striation information may lead to sub-optimal extraction and therefore may deteriorate the correlation result. In this paper, an edge detection algorithm and selection process are used together to locate the edge points of all tool-mark features and filter out those not corresponding to striation marks. Edge points of the resulting striation marks are reserved and expanded to generate a mask image. By imposing the mask image on the topography image, the weakly striated area(s) are removed from the expressed profile extraction. Using this method, 48 bullets fired from 12 gun barrels of six manufacturers resulted in a higher matching rate than previous studies.