Wireless Image Sensor Networks (WISN) where sensor nodes are equipped with miniaturized visual CMOS cameras to provide visual information is a promising technology for situation awareness, search&rescue or intrusion detection applications. In this paper, we present an off-the-shelf image sensor based on Arduino boards with a CMOS uCamII camera. The image sensor works with raw 128x128 image, implements an image change detection mechanism based on simple-differencing technique and integrates a packet losstolerant image compression technique that can run on very limited memory platforms. We detail the performance and energy consumption measures of the various image platforms and highlight how both medium-end and low-end platforms can be supported. From the single-camera system, we describe the extension to a multi-camera system which provides omnidirectional sensing at a very lost cost for largescale deployment.