Automated system and method for tool mark analysis
View Patent ↗An automated system for tool mark analysis includes an acquisition mechanism for acquiring 3D data of tool marks left on the surfaces of specimens by tools, a signature generation module for generating tool mark signatures from the acquired data, an analysis unit for comparing pairs of the tool mark signatures to obtain a numerical similarity value reflective of their similarity, and a database in which the similarity values are stored. An automated method for tool mark analysis involves the steps of acquiring the 3D data of tool marks, generating the tool mark signature, statistically evaluating pairs of the tool mark signatures, and providing a numerical similarity value for each pair of tool mark signatures reflecting the degree of similarity between the tool mark signatures of each pair. The method is carried out with automated assistance of a computer.
1. An automated system for tool mark analysis comprising
an acquisition mechanism for acquiring 3D data of tool marks left on the surfaces of specimens by tools operating on the specimens, said acquisition mechanism being capable of acquiring 3D from impressed tool marks and from striated tool marks;
a signature generation module for generating tool mark signatures from the acquired data corresponding to the respective tool marks;
an analysis unit for statistically evaluating pairs of the tool mark signatures for impressed tool marks in relation to one another and for statistically evaluating pairs of the tool mark signatures for striated tool marks in relation to one another, said analysis unit being capable of computing a numerical similarity value for each pair of tool mark signatures reflecting the degree of similarity between the tool mark signatures of each pair, said analysis unit employing a frequency domain-based similarity metric to compute a similarity value for each pair of tool mark signatures for impressed tool marks and employing a relative distance similarity metric to compute a similarity value for each pair of tool mark signatures for striated tool marks; and
a database in which said similarity values are stored.
2. The automated system for tool mark analysis recited in claim 1 wherein said database comprises questioned similarity values associated with questioned tool mark signatures of unknown origin and reference similarity values associated with control tool mark signatures of known origin corresponding to control specimens.
3. The automated system for tool mark analysis recited in claim 2 wherein said analysis unit is capable of performing a statistically based classification for the questioned similarity values against the reference similarity values.
4. An automated system for tool mark analysis comprising
an acquisition mechanism for acquiring 3D data of tool marks left on the surfaces of specimens by tools operating on the specimens, said acquisition mechanism including a 2D imaging mechanism for identifying regions of interest on the surfaces of the specimens for 3D data acquisition;
a signature generation module for generating tool mark signatures from the acquired data corresponding to the respective tool marks;
an analysis unit for statistically evaluating pairs of the tool mark signatures in relation to one another, said analysis unit being capable of computing a numerical similarity value for each pair of tool mark signatures reflecting the decree of similarity between the tool mark signatures of each pair; and
a database in which said similarity values are stored.
5. An automated system for tool mark analysis comprising
an acquisition mechanism for acquiring 3D data of tool marks left on the surfaces of specimens by tools operating on the specimens;
a signature generation module for generating tool mark signatures from the acquired data corresponding to the respective tool marks;
a data pre-processing module for operating on the acquired 3D data prior to signature generation, said data pre-processing module being capable of decimating the acquired 3D data, identifying unreliable data points in the acquired 3D data, replacing the unreliable data points with interpolated values, and identifying sections of the data of greatest reliability to obtain pre-processed data for each tool mark;
an analysis unit for statistically evaluating pairs of the tool mark signatures in relation to one another, said analysis unit being capable of computing a numerical similarity value for each pair of tool mark signatures reflecting the degree of similarity between the tool mark signatures of each pair; and
a database in which said similarity values are stored.
6. The automated system for tool mark analysis recited in claim 5 and further comprising a normalization module for operating on the pre-processed data prior to signature generation, said normalization module being capable of transforming the pre-processed data for each tool mark into a Cartesian coordinate representation and normalizing the Cartesian coordinate representation with respect to a reference surface.
7. The automated system for tool mark analysis recited in claim 6 wherein said normalization module is capable of applying a second-order leveling algorithm to the pre-processed data to obtain second-order leveled data for each tool mark, and applying a first-order leveling algorithm to the second-order leveled data to obtain first-order leveled data for each tool mark.
8. The automated system for tool mark analysis recited in claim 7 wherein the tool marks are striated tool marks and said signature generation module is capable of histogram equalizing the normalized data to obtain histogram equalized data, estimating local gradients for the histogram equalized data for each tool mark, identifying the dominant gradient direction for each tool mark from the corresponding local gradients, identifying the direction of the striations for each tool mark, projecting each tool mark onto a plane perpendicular to the direction of its striations to obtain a cross-sectional profile for each tool mark, and filtering the profiles to obtain the tool mark signatures for the tool marks.
9. The automated system for tool mark analysis recited in claim 8 wherein the similarity values computed by said analysis unit are based on relative distance similarity metrics.
10. The automated system for tool mark analysis recited in claim 7 wherein the tool marks are impressed tool marks and said signature generation module is capable of creating a data set from the normalized data for each tool mark contained in a two-dimensional data set to obtain the tool mark signature for each tool mark.
11. The automated system for tool mark analysis recited in claim 10 wherein the similarity values computed by said analysis unit are based on improved fast normal cross correlations.
12. The automated system for tool mark analysis recited in claim 6 and further including a calibration module for calculating calibration parameters associated with misalignments of the acquisition mechanism for use by said normalization module to refine the acquired data to compensate for the misalignments.
13. An automated method for tool mark analysis comprising the steps of
acquiring 3D data of tool marks left on the surfaces of specimens by tools operating on the specimens;
generating tool mark signatures for the respective tool marks from the acquired 3D data via a computer;
statistically evaluating pairs of the tool mark signatures in relation to one another via the computer, wherein said step of statistically evaluating includes implementing a frequency domain-based similarity metric where the pairs of tool mark signatures are for impressed tool marks and said step of statistically evaluating includes implementing a relative distance similarity metric where the pairs of tool mark signatures are for striated tool marks; and
providing a numerical similarity value via the computer for each pair of tool mark signatures reflecting the degree of similarity between the tool mark signatures of each pair.
14. An automated method for tool mark analysis comprising the steps of
acquiring 3D data of tool marks left on the surfaces of specimens by tools operating on the specimens;
preprocessing the acquired data to eliminate unreliable data points;
normalizing the preprocessed data to eliminate extraneous artifacts;
generating tool mark signatures for the respective tool marks from the acquired 3D data via a computer subsequent to said steps of preprocessing and normalizing;
statistically evaluating pairs of the tool mark signatures in relation to one another via the computer; and
providing a numerical similarity value via the computer for each pair of tool mark signatures reflecting the decree of similarity between the tool mark signatures of each pair.
15. The automated method for tool mark analysis recited in claim 14 wherein said step of preprocessing includes decimating the data, identifying unreliable data points, replacing unreliable data points with interpolated values, and identifying the most reliable section of the data.
16. The automated method for tool mark analysis recited in claim 14 wherein said step of normalizing includes transforming the preprocessed data into a Cartesian coordinate representation and normalizing the Cartesian coordinate representation with respect to a reference surface.
17. The automated method for tool mark analysis recited in claim 16 wherein said step of normalizing includes second order leveling of the data and first order leveling of the data.
18. The automated method for tool mark analysis recited in claim 17 wherein the tool marks are impressed tool marks and said step of generating includes filtering the normalized data.
19. The automated method for tool mark analysis recited in claim 16 wherein the tool marks are striated tool marks and said step of generating includes histogram equalizing the normalized data, estimating local gradients for every point of the histogram equalized data, identifying the dominant gradient direction, identifying the direction of the striations of the tool mark, obtaining a cross-sectional profile of the tool mark, and filtering the profile.