An optimal matched filter for target detection in images distorted by noise

To analyse automatically remote sensing images, many pattern recognition methods have been developed. Among the former, quote the segmentation that consists in recognition of textures in images or again the target detection whose the goal is to find a reference target in an observation sensor. This last is the object of this paper. The authors define a detection algorithm independent of the position of the sensor as compared to the target. They investigate how the optimal filter depends on the transformations, the prototype pattern and sensor properties, and they give explicit examples of optimal filter functions for active and passive sensors.