Examination of Firearms Review: 2007 to 2010 - Interpol

system (GIS), and a spatial-temporal algorithm developed for this project. The
results show that by using distance and time constraints interactively, the number
of candidate shoeprints that can implicate one or few suspects can be
substantially reduced. The study concludes that the use of space-time and other
ancillary information within a geographic information system can be quite helpful
for forensic investigation.
Many authors have recently proposed systems for automated recognition,
classification and retrieval of shoeprint patterns. Al-Garni and Hamiane (53)
developed an efficient automatic shoeprint retrieval system based on Hu’s moment
invariants. The performance of this algorithm is not significantly affected by
decreasing the image resolution. It is also shown that the optimal performance of
the proposed method is attained for rotated images. However, as stated by the
authors, this system is suitable only for comparing suspect outsole patterns, and
not partial shoeprints.
Dardi et al (54) presented an image retrieval algorithm which combines the
information of the phase of the Fourier transform of the shoe mark images with the
power spectral density of the Fourier transform calculated on their Mahalanobis
map. Different from other published studies, the algorithm performance here is
tested on real shoe marks from crime scenes. The proposed method is compared
with other studies and some preprocessing operators are also introduced and
selected to reduce noise and enhance the matching probability.
Pavlou and Allinson (55) developed an automated system for shoe model
identification from outsole impressions taken directly from the suspect's shoes that
can provide timely information while a suspect is in custody. The underlying
methodology is based on large numbers of localized features located using
maximally stable extermal region (MSER) feature detectors. These features are
transformed into robust scale invariant feature transform (SIFT) descriptors with
the ranked correspondence between footwear patterns obtained through the
application of modified constrained spectral correspondence methods. The
effectiveness of this approach is illustrated for a reference dataset of 374 different
shoe model patterns, from which 87% first-rank performance and 92% top-eight
rank performance are achieved. These authors were also involved in the
development of the Immersive Forensics Ltd. (UK) Latent Image Markup and
Analysis (LIMA) system, designed to provide a unified and intuitive environment
for the treatment of forensic images, including shoeprint and tire track impressions
(56).
A technique for automatic shoeprint matching, using multiresolution Gabor feature
map, was presented by Patil and Kulkarni (57). Gabor transform has been used to
extract genuine textural features in a shoeprint image. The proposed technique is
invariant to variations in intensity and rotation, and performs better compared to
results obtained using power spectral density (PSD) features for full print images
with rotation, intensity and mixed attacks at all the ranks. Performance of the
algorithm has been evaluated in terms of recognition rate and cumulative match
score for full prints and partial prints. The method was found to be robust to
Gaussian white noise and salt–pepper noise.
62