Abstract. Photographs of the back surface of the Turin Shroud were analysed to verify the existence of a double body image of a man. The body image is very faint and the background not uniform; i.e., the signal-to-noise ratio is lower than one. Therefore, image processing, developed ad hoc, was necessary to highlight body features. This was based on convolution with Gaussian filters, summation of images, and filtering in spatial frequency by direct and inverse bidimensional Fourier transformations. Body features were identified by template matching. The face and probably also the hands are visible on the back of the Turin Shroud, but not features related to the dorsal image.

Keywords: image processing, very low signal-to-noise ratio, fast Fourier transform, convolution, Turin Shroud

Print publication: Issue 6 (June 2004)
Received 13 October 2003, accepted for publication 12 March 2004
Published 13 April 2004