Energy aware multi-object detection method in visual sensor network

Nowadays, multi-object detection is one of the most important research issues in visual sensor networks. According to literature, many methods have been proposed for object detection. However, in most of the traditional works, preprocessing methods in camera nodes cause network to be faced with large amount of data when there are several objects in camera node field-of-views. In other words, conventional works can be suitable for single object detection. Therefore, in this paper, we propose an efficient preprocessing method in camera nodes for multi-object detection which is called Bounding box Boosting-based Face Detection (BBOD). In this method, first off, a background subtraction technique is used. Then the rows and columns of the objects bounding box are determined by a number of consecutive differences. After that, the bounding box of the all objects delivers to the Boosting-based face detection algorithm for extracting the faces of objects, and their boundaries. Finally, the detected faces are sent to the base station. The simulation results demonstrate that BBOD method injects less traffic to the network and saves camera nodes energy. Consequently, BBOD method increases the network lifetime in comparison with state-of-the-art algorithms.