Mathematical Morphology for Applications to Sensor Networks

Morphological image processing is a type of processing based on the theory of mathematical morphology, by which the spatial forms or the structures of objects in an image are modified. This objects-oriented image processing is advantageous for a sensor to make a decision, although is not good enough for visual aesthetics (or entertainment) purposes. Therefore, morphological image processing is suitable for sensing applications. It is noted that there are two special features associated with the video sensor networks: 1) limited resource and 2) the nonvisual aesthetics (or not for entertainment) purpose. Therefore, an approach of particular image processing on the sensor network is proposed in this paper. In this approach, the image is compressed using the quad tree data structure and is also processed in this structure using morphological operations to achieve the sensing purpose. Both the speed and the compression efficiencies of this proposed approach are verified from the sensor's perspective in our experiments.