Resource-efficient portable video communication system design for wildlife monitoring and interaction tracking

Abstract

In this research, we focus on algorithm development and system design for resource-efficient portable video communication system design and their application in wildlife monitoring and interaction tracking. The capability of seeing what an animal sees in the field is very important for wildlife activity monitoring and research. We design an integrated video and sensor system, called DeerCam and mount it on free-ranging animals so as to collect important video and sensor data about their activities in the field. From the video and sensor data collected by DeerCam, wildlife researchers will be able to extract a wealth of sciatic data for studying the behavior patterns of wildlife species and understanding the dynamic of wildlife systems. In this dissertation, we focus on the following four tightly coupled research issues: (1) Energy minimization. We develop joint power-rate-distortion (P-R-D) methods and algorithms for complexity control and energy minimization of portable video encoders. (2) Intelligent resource allocation and utility maximization. We develop methods to maximize the utility function under resource constraints. (3) Efficient image encoder. In DeerCam, a significant amount of video frames and image regions are encoded with the INTRA mode. We explore various image compression approaches to efficiently encode those frames and image regions. (4) Animal interaction detection for event-driven wildlife monitoring. We develop an animal interaction detection method using supervised learning methods to significantly reduce the amount of video data to be encoded and provide important reference for wildlife behavior analysis for wildlife researchers.