Feature Detection of Man-Made Objects Using Fractal Dimension

This project is intended to purpose a fractal dimension based object detection approach for man-made objects in the ultra-wideband foliage-penetrating (FOPEN) synthetic aperture radar (SAR) imagery. Since man-made objects are significantly different from natural ones from the point of view of fractal dimension, fractal dimension estimates can be used to characterize images and distinguish man-made objects from natural ones efficiently. In this project, natural scene and man-made object modeling with fractal geometric techniques will be pursued with an in-depth investigation of various fractal dimension definitions and their applicability to background characterization and object detection. Wavelet transforms will be utilized as the computational tools. Fractal dimension estimation algorithms will be implemented and compared. An object detection software which implements the fractal-based object detection algorithms will also be developed. In Phase II, the validated algorithms will be implemented in the hardware environment and verified through the prototype system.