Mine Detection Based on Adaboost and Polynomial Image Decomposition

摘要

In this paper, we propose a new method for underwater mine detection. This detection strategy is based on the use of the Adaboost algorithm with a Polynomial Image Decomposition (PID). PID splits a given image into two components the geometrical component (cartoon) and the textured one (small scale). This decomposition is based on the use of a polynomial transform. The use of PID reduces the noise and turbidity of underwater images, which results a consequent improvements on the visibility of underwater objects. As a result, our detector achieves a high detection rate and good efficiency. It also shows better performance against the use of a simple adaboost algorithm for underwater mine detection.