Міжвідомчий Науково-методичний збірник
"Криміналістика і судова експертиза"
ISSN: 0130-2655
PDF A2fc25cb69 280 289 Завантажень: 2, розмiр: 863.6 KB

V. Bondar


Pre-trial investigation and trial of criminal offenses committed with the use of firearms, requires solving a wide range of interrelated tasks within the framework of the use of expert technologies. One of them is the identification of firearms in the footsteps of shooting cartridges and bullets, which is carried out during forensic examinations. For this purposes are used automated ballistic identification systems.

The actual problem of automatic weapon identification is the development of effective algorithms for conducting checks on arrays that contain tens of thousands of digital images of the same type of traces of strikers.

There is a need in ABIS to process binary images. In such systems, high accuracy and speed of implementation of algorithms are required with minimal memory costs. Traditional methods of morphological transformations become ineffective in this case. In automated identification systems, the most acceptable are the geometric features of objects:

1. the area and perimeter of the image of the object;

2. the dimensions of inscribed and described simple geometric shapes (circles, rectangles, triangles, etc.);

3. number and relative position of angles;

4. moments of inertia of images of objects.

The individual signs of firearms that are reflected in the marks of the striker can be described with the help of various descriptors — characteristics that are uniquely determined. Descriptors that do not depend on the orientation of the image: area, perimeter, moments of inertia. These descriptors form a state vector of signs. Cluster analysis allows, on the basis of a comparison of such vectors, to draw conclusions about the similarity of traces.

The further direction of improving the use of ABIS, which will contribute to a more effective search for objects, is the development of the ideology of a single classifier for balls and cartridges, as well as traces remaining on their surfaces with parts of weapons and building on this basis an automated search system for weapons. To ensure the possibility of the formation by different researchers of a unified database of the search system, it is necessary to unify the technique for measuring trace parameters in accordance with the specified single classifier. Further improvement of computer technology will allow applying more sophisticated mathematical methods for processing trace information and make a transition to work with threedimensional characteristics of tracks. It is these changes that should contribute to the creation of a fully automatic forensic-ballistic identification system capable of solving not only technical but also analytical tasks.