Publication (Phd Thesis)

Contribución a la detección de objetivos específicos para aplicaciones de vigilancia con realidad aumentada

Cruz, Henry Omar

Abstract:

In this Doctoral Thesis, deficiencies have been identified in video surveillance systems regarding the effective and efficient detection of objects and regions with irregular shapes. The effectiveness is related to the capacity to detect specific objectives precisely, while efficiency is related to the possibility to present this information in real time to the user. Different proposals based on a methodological strategy of the treatment of color information are formulated in this work. This methodology contributes to the improvement of detections of specific targets with the mentioned irregular shapes, focused on resolving problems of environmental surveillance. In this context, the following three stages are presented: the extraction of the ICE, the calculation and application of the threshold TICG, applied to obtain the binary image of the detection and finally, the use of the relevant information to generate Augmented Reality. In the first stage, two color indices are proposed, the Forest Fire Detection Index (FFDI) and the Non Forest Detection Index (NFDI), which represent the optimal thresholds of for detecting forest fires and non-forested areas respectively. The level of precision reached through these indices is almost 96%, which is very high in comparison with other methods. They also have been proved to be applicable to real time applications. The capabilities and potentials found in the algorithms implemented throughout the proposed method allow the use of these algorithms as well as in traditional permanently installed systems as in mobile surveillance systems like drones. In this way, the change of the traditional video surveillance approach to a new scenario that guarantees mobility, flexibility and the extension of the coverage of surveillance is possible. In addition, two prototypes have been developed in this Thesis. Results verified the effectiveness and efficiency of the proposals and showed the possibilities to obtain and present virtual information from intrinsic information of the detected regions of interest.

Research areas:

• Image and Video Technologies

Year:

2017

Type of Publication:

Phd Thesis

Type of Publication:

PhD Thesis