Detecting External Disturbances on the Camera Lens in Wireless Multimedia Sensor Networks

Abstract

Assessing the quality of images acquired by nodes of a wireless multimedia sensor network (WMSN) is a critical issue, particularly in outdoor applications where external disturbances such as the presence of water, dust, snow, or tampering on the camera lens may seriously corrupt the acquired images. In this paper, we address the problem of determining when the lens of a microcamera mounted on a WMSN node is affected by an external disturbance that produces blurred or obfuscated images. We show that such a problem can be solved by considering change detection tests applied to a measure of the blur intensity. Interestingly, both measuring the blur and the change detection test do not require any assumption about the content of the observed scene, its nature, or the characteristics of the blur. The proposed methodology is thus flexible and effective for a large class of monitoring applications. In particular, it is attractive when nodes cannot continuously acquire images (e.g., because of energy limitations) and background subtraction methods are not feasible. Two solutions, with different hardware costs, have been designed, which can detect disturbances on the camera lens at either/both the node or/and the remote station level.