Détection radar utilisant les fractals et détection des cibles dans des images SAR utilisant des algorithmes de reconstruction des images dans un bruit non gaussien.

Abstract

The research work on radar detection has been widely investigated during the last decades. For this, several techniques have been developed to analyze and improve radar detection. However, the difficulty that arises in radar detection is to find an algorithm which adapts to a variety of environments encountered in practice. For this, it is always necessary to develop novel methods. An important concept which is in conjunction with the geometrical properties of an object is the fractal geometry. This geometry, which describes well the complex and irregular objects with its important parameter “the fractal dimension”, measures the degree of complexity of the structure considered and was used in radar detection. As part of our thesis, we propose a new radar detector based on the use of the fractal dimension, estimated by the method of box counting and adapted to all types of clutter in order to achieve the detection of radar signal in sea and ground clutter for synthetic and real data. Since its appearance, the radar imagery was subject to many studies, as well on the level of acquisition as the image processing rebuilt in order to improve the quality of information obtained. One of the advanced imaging radar technique is synthetic aperture radar (SAR) with its two types of configurations monostatic and bistatic. The process of generating a SAR image is undertaken via the use of signal processing techniques to form the image from raw data. Indeed, multiple image formation processes have been developed for monostatic SAR. In this work, we use three algorithms for generating the bistatic SAR images: Matched Filtering Algorithm (MFA), Back Projection Algorithm (BPA) and Polar Format Algorithm (PFA). We study the performance of these algorithms on two types of clutter; K and Weibull (real and complex).