Riconoscimento di facce in un sistema di visione distribuito

This thesis has been carried out within a project at IAS-Lab (Intelligent Autonomous Systems Lab), University of Padua. The project aims to create an autonomous system for distributed video surveillance. The thesis had as main objective the creation of a face detection and recognition system in a distributed sensors network. In particular, we wanted to create a system capable of processing video streams from sensors installed in different locations within a local network, in order to extract from them the images of the people faces in a room and then send the selected images to a central server, instructed to carry out the recognition. The main purpose of this system is to recognize people in the room even if they are not fully visible from some sensors, thus exploiting the distribution of the various cameras in the room. This thesis is mainly based on two aspects: - the creation of a reference model of the room in which the sensors are common to all of them; - the use of a middleware, called NMM, to create a network of nodes which distributes people and face detection and face recognition within a Local Area Network. In this thesis we will use the Viola-Jones algorithm to perform the detection of people and faces and the Eigenfaces algorithm for the recognition of the latter; we will also analyze the performance of the algorithm for recognition and propose some improvements aimed to increase the overall performance of the system