Gamma-backscattering imaging: evaluating the size of circular and ellipsoidal objects
W. Blanco 1,2, E. Merchan 1,2, F. Cristancho 1*, J. Gerl 3
1 Universidad Nacional de Colombia, Bogota, Colombia
2 Centro Internacional de Fisica, Bogota, Colombia
3 Helmholtzzentrum GSI GmbH, Darmstadt, Germany
The feasibility of producing images with backscattered gamma rays has already been experimentally demonstrated [1, 2]. One of the two 511 keV gamma rays of the positronic decay of 22Na goes to a pixelated detector whose register is activated if the other gamma ray, after being Compton-backscattered is detected by a scintillator some centimeters above the surface of the inspected soil.
Objects of various elemental compositions have been tested in soil of different types. Images of various qualities regarding contrast and distinguishibility of the object have been obtained. Computing techniques to improve the image quality and to evaluate geometrical parameters like shape and size are needed. We develop a procedure in which edge detection is performed by finding the regions of largest values (thresholding) of the gradient. This procedure defines a set of points on a plane that, in general, can thenbe tested with pattern recognition routines. We use here the Chernov-Lessort  and Modified Hough Transform , to obtain the radius of circular-haped objects, and the Least Squares Orthogonal Distance method  for the axes length of elliptical shapes. The different possibilities that each method offers in the context of gamma backscattering are analyzed. It is shown that no unique method is optimum, instead a judicious combination of the three presented here may be the best choice in practice.
 J. Gerl, Nucl. Phys. A752 (2005), 688c.
 E. Fajardo et al., AIP Conf. Proc. 1265 (2010), 449.
 N. Chernov and C. Lessort, J. Math. Imaging Vision, 23 (2005), 239.
 X. Cao and F. Deravi, Proc. 3rd Int. IEEE Conf. Comp. Vision, p. 744 (1990).
 Sung Joon Ahn, Least squares orthogonal distance fitting of curves and surfaces in space,