Content-based data prioritization for fast disaster images collection in delay tolerant network

When a large-scale disaster strikes, situational awareness over the disaster area is important because first responders need timely and accurate information for disaster relief. However, the communications infrastructure is usually unavailable thus a mobile ad-hoc network is utilized to gather critical information by using the mobile phones of people in the disaster zone as sensing nodes. In this paper, a content-based image prioritization technique in delay tolerant network is proposed to achieve fast collection of images with high priorities. Geotagged disaster images are collected by the users and are assigned priorities based on their content through image processing using OpenCV. A preinstalled mobile application on the users´ phones detects critical content (e.g., fires, road blocks) in the images. The detected images with critical content are sent over the network faster than images with noncritical content in order to address the critical events immediately. Through testing and computer simulation, we confirmed that the proposed method with image prioritization achieved a higher percentage of received images with critical content corresponding to a smaller delivery latency than without image prioritization.