IMPLEMENTATION OF AN ALGORITHM FOR DETECTING PATTERNS IN DEFOCUSED SCENES USING BINARY JOINT TRANSFORM CORRELATION

An original method to detect images in defocused scenes by means of a binary joint transform correlation is presented. Defocused images are described in frequency domain as the product between the Fourier trans form of the in-focus scene by the optical transfer function of the ima ging lens. This function, which presents positive and negative values, introduces sign errors in the joint power spectrum. The goal of this study is to detect the sign of the optical transfer function starting from the defocused image by means of a digital algorithm. Information provided by the algorithm allows to compensate the sign errors in the binary joint power spectrum and to obtain invariant correlations with respect to defocusing. Theoretical analysis and digital experiments to test the procedure are provided.