In this article, a parameter called striation density is proposed for determining and predicting bullet identifiability.
Automated firearms identification systems will correlate a reference bullet with all evidence bullets without a selection procedure to exclude the bullets having insufficient bullet identifying signature. Correlations that include such bullets increase the workload and may affect the correlation accuracy. After image preprocessing, the procedure described in this article uses edge detection and filtering techniques to extract the edges of striation marks. The resulting binary image distinctly shows the amount and distribution of striation marks. Then striation density is calculated for determining the quality of images. In the experiment, striation densities for 6 lands of 48 bullets fired from 12 gun barrels of 6 manufactures are calculated. Statistical results show strong relation between striation density and identification rate. It can provide firearms identification systems with a quantitative criterion to assess whether there are sufficient striae for reliable bullet identification. (Published Abstract)