Reducing the spatial resolution range of neutron radiographs cast by thick objects

The quality of a neutron radiograph is strongly dependent upon the features of the acquisition system. Most of them, such as detector resolution, electronic noise and statistical fluctuation can hardly be improved. Yet, a main parameter ruling the image spatial resolution, namely the L/D ratio of the system can be increased simply by lengthening the source-detector clearance. Such an option eventually may not be feasible due to neutron flux decreasing or engineering constraints. Under this circumstance, a radiograph improvement is only possible by some kind of after-acquisition procedure capable to retrieve, at least partially, the information concealed by the degradation process. Since the spoiling agent tied to the L/D has a systematic character, its impact can be reduced by an unfolding procedure such as Richardson-Lucy algorithm. However, that agent should be fully characterized and furnished to the algorithm as a Point Spread Function - PSF unfolding function. A main drawback of unfolding algorithms like Richardson-Lucy is that the PSF should be fixed, i.e., it assumes a certain constant image spatial resolution, rather than a variable one as actually occurs for thick objects. This work presents a methodology to minimize this difficulty by making all planes of the inspected object to cast a resolution within the shorter gap comprised between the object central plane and the detector. The image can then be unfolded with a lower resolution within a tighter range, yielding a better quality. The process is performed with two radiographs, where one of them is acquired with the object turned by 180° on its vertical axis with regard to the other. After a mirroring of one of them about its vertical axis, the images are added. As the resolution increases linearly with the object-detector gap, it would remain always lower than that of the central one. Therefore, the overall resolution of the composite radiograph is enhanced. A further improvement can then be achieved through an efficient unfolding since the object has been virtually shrunk along the neutron path.