Automatic Bullet Signature Identification System Based on Topography Measurements and Edge Detection Analysis
Wei Chu, John Song and Theodore Vorburger
Precision Engineering Division, MEL, NIST Gaithersburg, MD 20899.
Firearms often leave unique, reproducible markings on the fired bullets and cartridge cases. The analysis and comparison of these features play an important role in forensic analysis for crime investigation. Bullet signatures are striation marks; a procedure for automated bullet signature identification is described based on topography measurements obtained with a confocal microscope and edge detection analysis of the image. After the image acquisition and preprocessing, an edge detection technique is used to extract the feature edges. Then a filtering process is performed to remove all edge elements irrelevant to the striation marks. The finally obtained binary image distinctly reflects the amount and distribution of striation marks on the imaged bullet land. Using this binary image as a mask image, the bad areas that not contain striation signature are excluded for generating the 2D correlation profile. Based on the values of the cross-correlation maximum, a list of top ranking matches against a ballistics signature database of bullets fired from the same model gun is developed. The correlation results show a significant improvement using the sequence of methods described here compared with a currently used commercial system.
Mentors Name: Theodore Vorburger
Division, Laboratory: Precision Engineering Division, MEL
Room, Building Mail stop: B248/220, MS 8212
Tel: (301) 975-6595
Fax: (301) 869-0822
Not Sigma Xi Member