IMPLEMENTING IMAGE PROCESSING ALGORITHMS ORIENTED TO DETECT EMERGENCY SITUATIONS

ALCALA HUB RESEARCH TOPIC

IMPLEMENTING IMAGE PROCESSING ALGORITHMS ORIENTED TO DETECT EMERGENCY SITUATIONS

Motivation

Within the PhD Hub project initiative we would like to setup different research/innovation groups that bring together lead researchers, PhD candidates, and business/industry representatives. Each one of these groups will have a particular agenda and will result in different cooperation proposals, eg. Mobility of PhDs, joint PhDs, short-term research collaboration, knowledge transfer, etc.

In order to spark this collaboration, the following key research area is proposed in the Alcala hub (phdhub.eu/alcala).

Introduction

The use of Video Surveillance cameras is already well deployed in many scenarios. It is very usual to find out many places with cameras to guarantee the security of people. Thus these systems are providing a way to monitor and control the movement people, traffic, air-planes, etc. Nevertheless, the main detected problem is the automatic interpretation of the situation that cameras are capturing. Nowadays, intelligent systems are giving a new trend to identify if the situation is according to the expected pattern or not. Two ways are the most usual: centralized or decentralized systems. The first one is oriented to high-performance systems where all information from the captured images is processed. The second one pursues the implementation of smart camera systems.

This proposal is based on the second approach applied to scenarios where the detection of emergencies, anomalous behaviour or predefined dangerous situation can be processed in a smart camera. One application of this idea could be applied to crowded areas such as airports or cruise ships.

Research Description

Based on the above description, an emerging research in the field of image processing is proposed to provide novel solutions to detect “strange behaviours or situations” applied to spaces with a large number of people staying in or moving around. The development of specific algorithms/systems to count people, to detect trapped people or even to create heat maps with information concerning the area covered where there are people.

One crucial point will be the use of smart cameras where the implementation of ad-hoc algorithms on board could be available. This is very important because several solutions cannot allow downloading ad-hoc algorithms/programs on board. Another point to take into consideration will be the processing time. The use of a decentralized solution has to save the total execution time in terms of people/situation detection. To get this feature, the use of smart cameras within a complex embedded systems based on FPGAs, SoCs or high-performance microprocessors applied to the field of image processing should be considered.

Regarding the situations to detect, it is relevant to create solutions oriented to areas with people circulating or staying around that area. One example could be the large corridors in the transit areas in the airports where the safety of the people is a priority. To have the possibility to forecast anomalous situations and sending an alarm to the corresponding node, which is linked with the safety authorities can reduce the action time for the emergency/rescue team.

Looking for cooperation

Companies and researchers interested in the use of this approach to real scenarios are invited to declare their interest to create a common research activity aiming at providing solutions to this challenge. Among others joint PhDs, short-term research projects, knowledge transfer activities between research groups and interested companies may be the avenues for structuring such research collaboration.

Researchers registered in the PhD Hub platform can provide information about previous work done in the area of image processing specifically oriented to people detection, identification, analysis of scenarios and configuration/development of smart camera systems.

Deadline: 2^nd of June of 2020