Object recognition in lake and estuary environments

WCU Author/Contributor (non-WCU co-authors, if there are any, appear on document)
A. J. Punch (Creator)
Western Carolina University (WCU )
Web Site: http://library.wcu.edu/
Brian Howell

Abstract: Traditionally, autonomous underwater vehicles employ multiple configurations of sensor payloads in order to accomplish a specific mission. Due to advances in imaging technology, imaging sonar arrays and optical imaging devices are among these payloads. Independent of mission specifics, the majority of imaging data is either stored onboard the vehicle or transmitted to a base station for later analysis. In either situation, there is limited local real time analysis and limited mission duration. One focus for increasing real time analysis is the reduction of image information. By using image processing techniques, such as edge detection, less relevant information can be eliminated while preserving important object features. This reduced object information is then used as inputs to a neural network. A neural network is a cognitive algorithm which has the ability to adapt to achieve desired tasks. These networks are able to generalize and make decisions based on partial or limited input information. The goal of this research is to create an autonomous in-situ recognition system for marine environments, specifically the processing and classification of object image data. Image information will be applied to a neural network approach to mimic higher order decision making in an artificial cognitive algorithm.

Additional Information

Language: English
Date: 2011
Edge Detection, Neural Network, Optical, Sonar
Optical pattern recognition
Robot vision
Ocean engineering
Underwater imaging systems

Email this document to