Design and development of energy efficient compressive sensing based applications for wvsn

Abstract

A Wireless Visual Sensor Network (WVSN) is a collection of smart camera nodes that can capture, process and transmit images. The use of camera poses many challenges on resources, battery power and life time of the network. The power of visual sensor gets depleted, when huge data are transmitted and it reduces the network lifespan too. As the WVSN processes the visual data, time consumption is increased for data transmission and the storage requirement is also greater. This is a serious issue because, the WVSN is severely constrained to energy and failing to provide energy efficient solutions, the purpose of WVSN gets deteriorated. Understanding the demand of energy efficiency in WVSN, this research work presents three different energy efficient solutions for WVSN. The energy efficiency is attained by enforcing three different policies. The initial phase of the research aims to segregate the forepart from the background, as the background information is always static. Additionally, the compressive measurements are computed in a dynamic fashion and the selected compressive measurements alone are transmitted. This idea ensures energy efficiency. The second phase of the research conceives an idea that uniform sampling is unnecessary, as it degrades the image quality. On the other hand, uniform sampling consumes more energy. This issue is addressed by presenting an adaptable compressive sensing scheme for WVSN, which conserves energy by focussing on the textural features of the captured visual data. This idea preserves energy, while enhancing the quality of an image. The final phase of the research presents a compressive sensing based multi-focus image fusion scheme for WVSN. Suppose, when the captured visual data is not clear, data from multiple visual sensors can be combined together for obtaining a better view. Hence, this phase presents an compressive sensing based multi-focus image fusion scheme based on contourlet and curvelet. The image is rebuilt by the Compressive Sampling Matching Pursuit (CoSaMP) algorithm. The performances of all the proposed approaches are analysed in terms of precision, recall, F-measure, time complexity and energy consumption. The performances of the proposed approaches are compared with the existing approaches and the proposed approaches serve better newline