Resumo

RODRIGUEZ ARIAS, Felipe et al. Software component for weapons recognition in X-ray images. Rev cuba cienc informat [online]. 2017, vol.11, n.2, pp. 162-174. ISSN 2227-1899.

ABSTRACT The x-ray images constitute an important tool for security applications, especially on sites exploration of packages in for searching dangerous objects is very important. According to experts, the inspection of baggage by using the exploration of package techniques present limitations for the terms of efficiency in recognition of dangerous objects among other factors. This is influenced -in some aspects- by the characteristics of the X-ray images, which are very different to the visible spectrum images, affecting the performance of the techniques of exploration packages. For this reason, the acquisition methods, as well as for improving, processing or recognition of objects in these images, are characterize with expensive operations producing high runtimes. The reduction of the runtimes of the security applications can be a target of high priority for systems supported by fast decisions, among others. In addition, parallel programming techniques have proven to be a viable alternative for solving of problems in computer vision. In this paper, we propose a software component written in C/C++ that describes a procedure based on the BoVW algorithm handguns detection. For implementing the proposal, we use techniques of parallel programming with OpenMP, decreasing in 93.36% from a previous implementation of the proposal in MATLAB. The classification method is also tested in a new database of X-ray images, which were obtained by using dual-energy inspection equipment.

Palavras-chave : X-ray images; Bag of Visual Words; parallel computing..

        · resumo em Espanhol     · texto em Espanhol     · Espanhol ( pdf )