An edit distance model that can be used for the approximate matching of contiguous and noncontiguous timed strings is presented the paper. The model extends the concept of the weighted stringedit distance by introducing timed edit operations and by making the edit costs time dependent. Special attention is paid to the timed null symbols that are associated with the timed insertions and deletions. The proposed original algorithm can be applied to general problems involving the comparison of time series of events and thus also for evaluating intelligent surveillance systems.
Person recognition using facial features extracted, for example, from mugshot images, represents one of the most established research fields of biometrics. However, due to the widespread use of webcams and mobile devices embedded with a camera, it is now possible to realize facial video recognition, rather than resorting to just still images. Due to these developments, we are currently witnessing a shift in the research focus of many research groups, which direct their efforts more towards videobased face recognition and away from recognition techniques relying solely on still images. The reason for this trend can be found in the advantages facial video recognition offers over still image recognition; these include the potential of boosting the system’s accuracy and robustness as well as deterring spoof attacks. This paper presents an evaluation of person identity verification using facial video data. It involves 18 systems made available by seven academic institutes such as IDIAP, the University of Surrey, the University of Ljubljana, etc. These systems provide for a diverse set of assumptions, including feature representation and preprocessing variations, allowing us to assess the effect of adverse conditions, usage of quality information, query selection, and template construction for videotovideo face authentication. The analysis and findings presented in the paper serve as a reference point for researchers working in the field of face recognition and offer an important insight into the characteristics of different face recognition techniques when applied to facial video data.
Face recognition systems exploiting Gabor filters are amongst the most robust and efficient face-based biometric systems in existence today. Commonly, these systems adopt a number of complex Gabor filters and use the magnitude responses of the filtering operation to derive useful features for the recognition task. Different from existing techniques that use Gabor filters for deriving facial features, we propose in this paper a novel feature extraction technique that does not rely solely on Gabor magnitude information but effectively uses the Gabor phase information as well. We show that our Gabor phase features contain complementary information to Gabor magnitude features and that the two feature types can easily be combined for enhanced face recognition performance. In the experimental section of the paper we demonstrate that the combined features ensure highly efficient and most of all robust recognition performance, which remains relatively stable regardless of the presence of external distortions caused, for example, by variable lighting, pose variations or partial occlusions of the facial area.
This was a research in the domain of environmental measurements modeling, where we developed a suitable model for the analysis of the key environmental parameters of the air quality, measured by particulate matter PM10 and black carbon (BC) concentrations. In this study the hourly derived day-to-day profiles of the PM10 and BC concentrations in the Port of Koper were examined over a one-year period. and investigated how the different diurnal profiles were distributed over this period. The data were analysed by clustering the days into groups according to the similarity of the BC and PM10 hourly derived profiles. This approach made it possible to investigate how many different typical day-profiles were present during the one-year period and to establish the relationships of the various typical day-profiles to seasonal and weather conditions, to working and non-working hours and days and also to port activities.
We consider similarity-based relational databases that allow to retrieve approximate or relevent data. We focus on the recently introduced relational algebra with similarities on D-relations, which are annotated with multi-dimensional similarity values with each dimension referring to a single attribute. The codomains of the annotated relations are De Morgan frames, and the annotations express the relevance of the tuples as answers to a similarity-based query. In this paper, we study Datalog programs on D-relations, with and without negation. We describe the least-fixpoint algorithm for safe and rectified Datalog programs on D-relations with finite support but without negative literals in the body. We further describe the perfect-minimal-fixpoint algorithm of a Datalog program on D-relations with finite support and negative literals in the body when rules are safe, rectified and stratified. We introduce the idea of controlling the calculation of the annotations such that the tuples that enter an IDB relation last will be announced less desirable than those that enter first. For this we define a damping function that augments/diminishes the individual annotations that contribute to the final annotations of tuples.