Firearm detection from bullet casing using retrographic elastomeric sensor

Fired bullet casings are a property that shows the distinctive of firearms. In case this feature is used in firearm detection, ballistic examinations with bullet casings can take a long time in laboratory environments. In this thesis, it is aimed to quickly detect firearms from bullet casings with a mobile system design independent of the laboratory environment. In order to convert the rough areas, which express the characteristic details on the back surface of the fired bullet case, into detailed data, micron precision image data from this surface is needed. It is impossible to obtain these necessary images with only light and a camera due to the glare on the metal surface. In order to solve this problem, inspired by the sense of touch, which is a biological feature, it is considered to design the imaging unit of the retrographic elastomeric sensor by using glazed elastomer and a microscopic camera, which will enable to show the roughness on the back surface of the bullet case at the micron level. It is assumed that the retrographic images to be obtained with this design unit can be quickly analyzed with a software developed based on deep learning, and firearm detection can be made from the bullet case. Platinum-based transparent silicone material can be briefly described as an elastomer. Elastomers are polymer-derived elastic materials and the manufacture of the elastomer is carried out by mixing the components called A, B and C in certain proportions. In order to obtain the elastomeric texture that will capture the best tactile sensation, it is necessary to achieve the appropriate mixing ratio of the components. Taguchi orthogonal array design was used for this. In the experiment, the appropriate material was determined by considering the criteria such as softness, strength, light transmittance. The retrographic images obtained with the designed system were converted into a data set form, and the distinctive feature of these data set images was used in deep learning methods, eventually thanks to these images, the detection of firearm from the bullet casing was successfully carried out.